No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Machine learning-based prediction of adverse events following an acute coronary syndrome (PRAISE): a modelling study of pooled datasets
Abstract
Background
The accuracy of current prediction tools for ischaemic and bleeding events after an acute coronary syndrome (ACS) remains insufficient for individualised patient management strategies. We developed a machine learning-based risk stratification model to predict all-cause death, recurrent acute myocardial infarction, and major bleeding after ACS.
Methods
Different machine learning models for the prediction of 1-year post-discharge all-cause death, myocardial infarction, and major bleeding (defined as Bleeding Academic Research Consortium type 3 or 5) were trained on a cohort of 19 826 adult patients with ACS (split into a training cohort [80%] and internal validation cohort [20%]) from the BleeMACS and RENAMI registries, which included patients across several continents. 25 clinical features routinely assessed at discharge were used to inform the models. The best-performing model for each study outcome (the PRAISE score) was tested in an external validation cohort of 3444 patients with ACS pooled from a randomised controlled trial and three prospective registries. Model performance was assessed according to a range of learning metrics including area under the receiver operating characteristic curve (AUC).
Findings
The PRAISE score showed an AUC of 0·82 (95% CI 0·78–0·85) in the internal validation cohort and 0·92 (0·90–0·93) in the external validation cohort for 1-year all-cause death; an AUC of 0·74 (0·70–0·78) in the internal validation cohort and 0·81 (0·76–0·85) in the external validation cohort for 1-year myocardial infarction; and an AUC of 0·70 (0·66–0·75) in the internal validation cohort and 0·86 (0·82–0·89) in the external validation cohort for 1-year major bleeding.
Interpretation
A machine learning-based approach for the identification of predictors of events after an ACS is feasible and effective. The PRAISE score showed accurate discriminative capabilities for the prediction of all-cause death, myocardial infarction, and major bleeding, and might be useful to guide clinical decision making.
Funding
None.
... Machine learning or other artificial intelligence methods could also be adopted in this integrative approach to include more risk factors, not just morphological and biomechanical factors to assess future plaque behaviors, particularly when a large volume of data is being analyzed [115][116][117] . Based on 19826 patients with ACS, PRAISE study group have demonstrated the feasibility and effectiveness of machine learning-based approaches to predicting all-cause death, recurrent acute myocardial infarction, and major bleeding after acute coronary syndrome 115 . ...
... Machine learning or other artificial intelligence methods could also be adopted in this integrative approach to include more risk factors, not just morphological and biomechanical factors to assess future plaque behaviors, particularly when a large volume of data is being analyzed [115][116][117] . Based on 19826 patients with ACS, PRAISE study group have demonstrated the feasibility and effectiveness of machine learning-based approaches to predicting all-cause death, recurrent acute myocardial infarction, and major bleeding after acute coronary syndrome 115 . Koloi et al. also shown that machine learning methods can accurately predict early-stage coronary artery disease using a set of clinical characteristics and routine laboratory markers 116 . ...
... Biomechanical modeling has been increasingly employed in the diagnosis, clinical decision-making, and treatment for patients with coronary artery disease 115,118 . A notable example is fractional flow reserve (FFR) in coronary atherosclerotic lesions, which reflects the pressure drop from the proximal aorta to the coronary segment distal to the lesion during maximal vasodilation 118,119 . ...
Coronary biomechanics including structural wall stress and strain on the vessel wall and flow wall shear stress on the endothelial surface play a vital role in plaque progression, rupture and erosion. This review summarizes recent advances in coronary intravascular imaging, image-based biomechanical modeling, and their applications in investigating possible biomechanical mechanisms of plaque rupture and erosion, and developing interventional therapies targeting unfavorable mechanical conditions for personalized treatment and precision medicine.
... Overall, 59 studies were included in the current systemic review from which 15 were on long-term mortality [20][21][22][23][24][25][26][27][28][29][30][31][32][33][34] (Fig. 1). Excluded articles in the full-text screening and reasons for exclusion are provided in the Supplementary Material 1. ...
... Supplementary Table S1-2 (Supplementary Material 2) demonstrate the general characteristics of the studies. In brief, fifteen articles assessed the performance of ML on long-term mortality of which seven (46%) were included in the meta-analysis [20,23,24,27,29,32,34]. 40% (6/15) of the studies did not report an event per variable (EPV) [22-24, 31, 33, 34], while in the others this figure was from 1.1 to 28.8 with only one study [26] having an EPV > 10. No studies used multiple imputations for handling the missing values and 53% (8/15) studies did not report their methods for missing data [22, 24-27, 29, 31, 32]. ...
... No studies used multiple imputations for handling the missing values and 53% (8/15) studies did not report their methods for missing data [22, 24-27, 29, 31, 32]. Only 40% (6/15) studies reported model calibration [20,[25][26][27][28][29], and 26% (4/15) had an external validation dataset [20,26,28,29]. ...
Introduction
Percutaneous coronary intervention (PCI) has been the main treatment of coronary artery disease (CAD). In this review, we aimed to compare the performance of machine learning (ML) vs. logistic regression (LR) models in predicting different outcomes after PCI.
Methods
Studies using ML or deep learning (DL) models to predict mortality, MACE, in-hospital bleeding, and acute kidney injury (AKI) after PCI or primary PCI were included. Articles were excluded if they did not provide a c-statistic, solely used ML models for feature selection, were not in English, or only used logistic or LASSO regression models. Best-performing ML and LR-based models (LR model or conventional risk score) from the same studies were pooled separately to directly compare the performance of ML versus LR. Risk of bias was assessed using the PROBAST and CHARMS checklists.
Results
A total of 59 studies were included. Meta-analysis showed that ML models resulted in a higher c-statistic compared to LR in long-term mortality (0.84 vs. 0.79, P-value = 0.178), short-term mortality (0.91 vs. 0.85, P = 0.149), bleeding (0.81 vs. 0.77 P = 0.261), acute kidney injury (AKI; 0.81 vs. 0.75, P = 0.373), and major adverse cardiac events (MACE; 0.85 vs. 0.75, P = 0.406). PROBAST analysis showed that 93% of long-term mortality, 70% of short-term mortality, 89% of bleeding, 69% of AKI, and 86% of MACE studies had a high risk of bias.
Conclusion
No statistical significance existed between ML and LR model. In addition, the high risk of bias in ML studies and complexity in interpretation undermines their validity and may impact their adaption in a clinical settings.
... In this study, we aimed to evaluate the association between the Hb/Cr ratio at hospital discharge and one-year clinical outcomes in patients with STEMI using data from the large, multinational PRAISE registry [18]. We explored the utility of this index as a potential surrogate marker of patient vulnerability and aimed to identify threshold values which may guide possible risk stratification strategies. ...
... This study is a retrospective observational analysis based on the PRAISE registry, a pooled dataset derived from several international registries and prospective cohorts of patients with ACS. The registry was originally constructed to develop and validate the PRAISE risk scores using machine learning models for the prediction of 1-year postdischarge all-cause mortality, myocardial infarction, and major bleeding following ACS [18]. It incorporates patient-level data collected from 2003 to 2019, from a total of six large-scale datasets: the BleeMACS registry (NCT02466854) [19,20], the RENAMI registry [19,20], the FRASER study (NCT02386124) [21], the Prospective Registry of Acute Coronary Syndromes in Ferrara (NCT02438085), the SECURITY randomized controlled trial [22], and the Clinical Governance in Patients with ACS project (NCT04255537). ...
... Full inclusion and exclusion criteria and methodological details have been published previously and are available on ClinicalTrials.gov [18][19][20][21][22][23][24]. ...
Background/Objectives: Anemia and renal impairment are key predictors of adverse outcomes in acute coronary syndromes (ACSs). The hemoglobin-to-creatinine (Hb/Cr) ratio combines these parameters into a simple index. This study aimed to evaluate its prognostic value at discharge in patients with ST-elevation myocardial infarction (STEMI). Methods: The primary endpoint was one-year all-cause mortality; secondary endpoints included major bleeding and the composite of all-cause mortality or reinfarction. Optimal Hb/Cr cut-off values were identified using Liu’s method. Multivariable logistic regression and propensity score matching were used to assess associations with outcomes. Results: We analyzed 11,236 STEMI patients from the PRAISE registry with available hemoglobin and creatinine values at discharge. The optimal cut-points were 13.68 for mortality and 14.42 for secondary endpoints. Patients were stratified into low (<13.68; 26.5%) and high (≥13.68; 73.5%) Hb/Cr groups. The low Hb/Cr group was older, had more comorbidities, and received less intensive therapy. At one year, low Hb/Cr patients had significantly higher rates of all-cause mortality (8.7% vs. 2.4%), major bleeding (5.0% vs. 2.4%), and the composite outcome (11.5% vs. 4.9%). In the multivariate logistic regression, the Hb/Cr ratio was inversely associated with all outcomes, namely all-cause mortality (odds ratio [OR] 0.94; 95% confidence interval [CI]: 0.92–0.96), major bleeding (OR 0.96; 95% CI: 0.94–0.97), and the composite endpoint (OR 0.93; 95% CI: 0.91–0.96). The Hb/Cr ratio outperformed hemoglobin and creatinine alone in predicting mortality (AUC 0.684 vs. 0.649 and 0.645; p < 0.001). Conclusions: The Hb/Cr ratio is independently associated with one-year adverse outcomes in STEMI and may serve as a simple marker of increased vulnerability. Prospective studies are needed to validate its clinical utility.
... Based on the above, the stratification of ACS patients for the risk of early developing VA or AF carries important prognostic and therapeutic implications [5,6]. The PRAISE (PRedicting with Artificial Intelligence riSk aftEr acute coronary syndrome) score [7] was recently designed to develop a machine learningbased risk stratification model able to predict the risk of allcause death, recurrent myocardial infarction (MI) and major bleeding after ACS and guide clinical decision making. To date, its role in predicting arrhythmic complications in ACS is unknown. ...
... In detail, the PRAISE data set [7,8] , prior MI, prior PCI, CABG, prior stroke, prior bleeding, malignancy, STEMI presentation, hemoglobin, and LVEF); 5 therapeutic variables (beta-blocker, angiotensin-converting enzyme inhibitor/ angiotensin-receptor blocker, statin, oral anticoagulation, and proton pump inhibitor); 2 angiographic variables (multivessel disease and complete revascularization); 2 procedural variables (vascular access and PCI with drug-eluting stent). The PRAISE score was able to separately predict the occurrence of three different outcomes 1 year after discharge: all-cause death, new event of MI and major bleeding (by 3 or 5 Bleeding Academic Research Consortium [BARC] definition) [9]. ...
... Machine learning algorithms, exploring high-dimensional and nonlinear relations among features, could represent a novel approach to the compelling requirement of a personalized risk assessment. Indeed, the PRAISE score, derived from a machine learning process, in the derivation and validation cohorts appeared to overperform for the stratification of the risk of adverse events compared to existing ischemic and bleeding risk scores [7,10]. Notably, in a recent work we found that the adoption of the PRAISE score was effective for addressing the issue of a tailored intensity of dual antiplatelet therapy after ACS based on patients' offsetting bleeding and ischemic profiles [8]. ...
Background
The PRAISE (PRedicting with Artificial Intelligence riSk aftEr acute coronary syndrome) score is a machine learning‐based model for predicting 1‐year adverse cardiovascular or bleeding events in patients with acute coronary syndrome (ACS). Its role in predicting arrhythmic complications in ACS remains unknown.
Methods
Atrial fibrillation (AF) and ventricular arrhythmias (VA) were recorded by continuous electrocardiographic monitoring until discharge in a cohort of 365 participants with ACS prospectively enrolled. We considered two separate timeframes for VA occurrence: ≤ 48 and > 48 h. The objective was to evaluate the ability of the PRAISE score to identify ACS patients at higher risk of in‐hospital arrhythmic complications.
Results
ROC curve analysis indicated a significant association between PRAISE score and risk of both AF (AUC 0.89, p = 0.0001; optimal cut‐off 5.77%) and VA (AUC 0.69, p = 0.0001; optimal cut‐off 2.17%). Based on these thresholds, high/low AF PRAISE score groups and high/low VA PRAISE score groups were created, respectively. Patients with a high AF PRAISE score more frequently developed in‐hospital AF (19% vs. 1%). Multivariate analysis showed a high AF PRAISE score risk as an independent predictor of AF (HR 4.30, p = 0.016). Patients with high VA PRAISE scores more frequently developed in‐hospital VA (25% vs. 8% for VA ≤ 48 h; 33% vs. 3% for VA > 48 h). Multivariate analysis demonstrated a high VA PRAISE score risk as an independent predictor of both VA ≤ 48 h (HR 2.48, p = 0.032) and VA > 48 h (HR 4.93, p = 0.014).
Conclusion
The PRAISE score has a comprehensive ability to identify with high specificity those patients at risk for arrhythmic events during hospitalization for ACS.
... Единственный алгоритм, сфокусированный на определение риска ИБС с учётом анемии, создан F. D'Ascenzo и соавт. [25]. Показатели красной крови оценивали как параметр прогноза исхода анемии только T. Ohara и соавт. ...
... D'Ascenzo и соавт.) [25]. Частота использования КТ для оценки исходов, прогнозируемых авторами статей, представлена в табл. 1. Оценка параметров прогнозирования исходов говорит о том, что преобладает фокус на «клинические» КТ -смерть, аритмия, кровотечение, а «морфологические» неблагоприятные КТ, к которым можно отнести гипертрофию миокарда и тромбоз стента, оцениваются лишь в 7,8% случаев суммарно. ...
... Однако этот показатель включают в анализ риска при ИБС лишь 10% исследователей. Алгоритмы T. Ohara и F. D'Ascenzo являются первыми опубликованными платформами прогнозирования для лиц с ИБС, учитывающими анемический синдром, при этом оба алгоритма создавались для пациентов с кровотечениями и/или c исходной анемией [24,25]. У пациентов без анемии или не имевших кровотечений никто из исследователей не включил анемию в число возможных исходов, которой следует прогнозировать. ...
Materials and methods. The PubMed and RSCI databases were analyzed covering the period of 2000-2023. 906 articles were found using the keywords “artificial intelligence”, “anemia”, coronary artery disease”, “hemoglobin”, “cardiac surgery”, of which 38 met the criteria for inclusion in the analysis. Results. In a number of countries around the world, artificial intelligence (AI) systems have now been created to predict the course of IHD, however, at the moment, data have been published on the single system with AI elements, presented by the developers of the University of Turin (Italy). It has the functionality of predicting the course of IHD and complications of invasive procedures for IHD against the background of anemic syndrome, based on the use of the HAS-BLEED scale. The increase in the number of CABG operations determines the importance of further research into their long-term results and the development of programs for their management, which will take into account such factors that are important for choosing a strategy for their management and the possibility of influencing the risks of an unfavorable prognosis. This review presents published data on the developed and used digital products based on artificial intelligence intended for the management of patients with coronary artery disease, including taking into account basic hematological parameters. Conclusion. Analysis of existing developed AI systems showed a focus on solving prognostic issues. At the same time, in our opinion, the range of analyzed parameters is not wide enough, in particular, taking into account the presence of anemia, which plays one of the key roles in modifying the risk of adverse outcomes (coronary deaths, repeated acute coronary events, progression of CHF).
... Indeed, while many AI algorithms and applications demonstrate high performance in retrospective analyses, they encounter challenges in generalizability when applied to diverse patient populations or challenging settings. 4,5 Indeed, in order to ensure robust clinical adoption, external validation in representative cohorts is essential, particularly in interventional cardiology where real-time decision-making and seamless articulation of materials are paramount. ...
... The clinical translation of improved predictive modeling into personalized decision-making remains a challenge. 5 While ML techniques can identify patterns linked to myocardial infarction, heart failure, and stroke, their reported predictive accuracy rarely translates to real-world settings due to overfitting, biased training data, and poor generalizability. 25 Similarly, claims on the favorable impact of AI-guided therapy on drug selection and adherence strategies are reassuring, but evidence of benefits over clinician-guided care remains limited. ...
Artificial intelligence (AI) is being intensively applied to cardiology, particularly in diagnostics, risk prediction, treatment planning, and invasive procedures. While AI-driven advancements have demonstrated promise, their real-world implementation remains constrained by critical challenges. Current AI applications, such as electrocardiogram interpretation and automated imaging analysis, have improved diagnostic accuracy and workflow efficiency, yet generalizability, regulatory hurdles, and integration into existing clinical workflows remain major obstacles. Algorithmic bias and the lack of explainable AI further complicate widespread adoption, potentially leading to disparities in healthcare outcomes. In interventional cardiology, robotic-assisted percutaneous coronary intervention has emerged as a technological innovation, but comparative clinical evidence supporting its superiority (or even non-inferiority) over conventional approaches is still limited. Additionally, AI-based decision support systems in high-risk cardiovascular procedures require rigorous validation to ensure safety and reliability. Ethical considerations, including patient data security and region-specific regulatory frameworks, also pose significant barriers. Addressing these challenges requires interdisciplinary collaboration, robust external validation, and the development of transparent, interpretable AI models. This review provides a critical appraisal of the current role of AI in cardiology, emphasizing both its potential and its limitations, and outlines future directions to facilitate its responsible integration into clinical practice.
... The data on the prognostic significance of severe coronary atherosclerotic burden on MACEs in the mid-term in ACS patients are less robust than in stable CAD. In previous studies of ACS, the relative independent prognostic significance of multivessel coronary artery disease was small [35]. In our study, three-vessel coronary artery disease was a predictor of MACEs, but its effect became insignificant when adjusted for other variables. ...
... Risk factor control is an important determinant of prognosis after ACS. In the BARI 2 D study of patients with stable CAD, risk factor control 1 year after inclusion in the study was closely associated with survival and cardiovascular outcomes [35]. Similar results were obtained in our study. ...
Due to the routine use of endovascular revascularization and improved medical therapy, the majority of acute coronary syndrome (ACS) cases now have an uncomplicated course. However, in spite of the currently accepted secondary prevention standards, the residual risk of remote major adverse cardiovascular events (MACEs) after ACS remains high. Ultrasound carotid/subclavian atherosclerotic plaque assessment may represent an alternative approach to estimate the MACE risk after ACS and to control the quality of secondary prevention. Aim: To find the most important clinical predictors of MACEs in contemporary patients with predominantly uncomplicated ACS treated according to the Guidelines, and to study the potential of the longitudinal assessment of quantitative and qualitative ultrasound carotid/subclavian atherosclerotic plaque characteristics for MACE prediction after ACS. Methods: Patients with ACS, obstructive coronary artery disease (CAD) confirmed by coronary angiography, and carotid/subclavian atherosclerotic plaque (AP) who underwent interventional treatment were prospectively enrolled. The exclusion criteria were as follows: death or significant bleeding at the time of index hospitalization; left ventricular ejection fraction (EF) <30%; and statin intolerance. The clinical variables potentially affecting cardiovascular prognosis after ACS as well as the quantitative and qualitative AP characteristics at baseline and 6 months after the index hospitalization were studied as potential MACE predictors. Results: A total of 411 primary patients with predominantly uncomplicated ACS were included; AP was detected in 343 of them (83%). The follow-up period duration was 450 [269; 634] days. MACEs occurred in 38 patients (11.8%): seven—cardiac death, twenty-five—unstable angina/acute myocardial infarction, and six—acute ischemic stroke. In multivariate regression analyses, the most important baseline predictors of MACEs were diabetes (HR 2.22, 95% CI 1.08–4.57); the decrease in EF by every 5% from 60% (HR 1.22, 95% CI 1.03–1.46); the Charlson comorbidity index (HR 1.24, 95% CI 1.05–1.48); the non-prescription of beta-blockers at discharge (HR 3.24, 95% CI 1.32–7.97); and a baseline standardized AP gray scale median (GSM) < 81 (HR 2.06, 95% CI 1.02–4.19). Among the predictors assessed at 6 months, after adjustment for other variables, only ≥ 3 uncorrected risk factors and standardized AP GSM < 81 (cut-off value) at 6 months were significant (HR 3.11, 95% CI 1.17–8.25 and HR 3.77, 95% CI 1.43–9.92, respectively) (for all HRs above, all p-values < 0.05; HR and 95% CI values varied minimally across regression models). The baseline quantitative carotid/subclavian AP characteristics and their 6-month longitudinal changes were not associated with MACEs. All predictors retained significance after the internal validation of the models, and models based on the baseline predictors also demonstrated good calibration; the latter were used to create MACE risk calculators. Conclusions: In typical contemporary patients with uncomplicated interventionally treated ACS, diabetes, decreased EF, Charlson comorbidity index, non-prescription of beta-blockers at discharge, and three or more uncontrolled risk factors after 6 months were the most important clinical predictors of MACEs. We also demonstrated that a lower value of AP GSM reflecting the plaque vulnerability, measured at baseline and after 6 months, was associated with an increased MACE risk; this effect was independent of clinical predictors and risk factor control. According to our knowledge, this is the first demonstration of the independent role of longitudinal carotid/subclavian AP GSM assessment in MACE prediction after ACS.
... The authors employed clinical parameters required for calculating the PRAISE score [1], a tool originally developed using machine learning to predict 1-year adverse cardiovascular and bleeding events following ACS [2]. However, additional potential predictors are known to influence arrhythmogenesis. ...
... Identifying modifiable risk factors provides actionable therapeutic targets to mitigate the incidence of AF and VAs post-ACS. For instance, anemia emerged as a significant predictor in the original PRAISE cohort, alongside age and left ventricular ejection fraction [2]. Notably, anemia is widely recognized as a contributor to the pathogenesis of AF and may represent a practical focus for intervention. ...
... Despite these insights, there remains a paucity of studies that systematically and dynamically monitor electrical impedance along meridian pathways. To address this gap, electronic devices were employed in the present study to measure bioelectrical impedance (BEI) along 12 pairs of meridian pathways in healthy individuals [15][16][17]. The primary objective was to characterize the variation patterns of electrical impedance across these pathways over time. ...
... Despite these insights, there is still a lack of systematic studies on dynamic electrical impedance along meridians. To address this gap, this study used electronic devices to measure bioelectrical impedance (BEI) along 12 pairs of meridian pathways in healthy individuals, with the goal of analyzing impedance variation patterns over time [15][16][17]. ...
Objective
This study aims to investigate the monthly variation patterns of bioelectrical impedance (BEI) along 24 meridian pathways in healthy individuals.
Methods
A cohort of 684 healthy middle-aged participants from North China was enrolled between July 1, 2017, and September 5, 2020. BEI measurements were consistently recorded along the 24 meridian pathways over the study period. The collected BEI data were subjected to statistical analysis, and line charts were constructed to depict the temporal variation patterns.
Results
Analysis revealed that BEI values along the 24 meridian pathways followed a normal distribution over a 12-month period. In the first group of meridians, which includes the lung, large intestine, heart, small intestine, pericardium, and triple-energizer meridians, significant monthly variations were observed. The second group, comprising the spleen, stomach, bladder, kidney, gallbladder, and liver meridians, exhibited marked differences primarily between March and April (P < 0.05), with a peak in April and relatively stable values thereafter. Synchronous BEI fluctuations were evident on the left and right sides of the body, and both groups of meridian pathways displayed similar variation patterns. These patterns largely corresponded to fluctuations observed in the spleen meridian.
Conclusion
The consistent monthly variation patterns in BEI along the 24 meridian pathways among healthy middle-aged individuals align with Traditional Chinese Medicine (TCM) concepts of meridians and collaterals. The spleen meridian, in particular, appears to play a crucial role in influencing these bioelectrical fluctuations, as posited in TCM theory. From a bioelectrical standpoint, this study provides empirical support for the potential existence and functionality of meridians and collaterals, offering a scientific perspective that complements ancient TCM principles.
... Recently, machine learning-based prediction models have emerged as potential tools for disease progression prediction [6,7]. These models are believed to better handle complex, highdimensional data relationships and more accurately reflect the associations between variables and outcomes compared to traditional linear models such as the Cox model. ...
... The C-index is a widely recognized metric that quantifies the ability of a model to predict outcomes. A model with an AUC greater than 0.75 is generally considered to exhibit excellent discrimination [7]. Calibration was appraised using the Brier score at the same time points; a Brier score of 0.25 or less signifies favorable model calibration [17]. ...
Idiopathic pulmonary fibrosis (IPF) is a chronic interstitial lung disease with a poor prognosis. Its non-specific clinical symptoms make accurate prediction of disease progression challenging. This study aimed to develop molecular-level prognostic models to personalize treatment strategies for IPF patients. Using transcriptome sequencing and clinical data from 176 IPF patients, we developed a Random Survival Forest (RSF) model through machine learning and bioinformatics techniques. The model demonstrated superior predictive accuracy and clinical utility, as shown by the concordance index (C-index), the area under the operating characteristic curve (AUC), Brief scores, and decision curve analysis (DCA) curves. Additionally, a novel prognostic staging system was introduced to stratify IPF patients into distinct risk groups, enabling individualized predictions. The model’s performance was validated using a bleomycin-induced pulmonary fibrosis mouse model. In conclusion, this study offers a new prognostic staging system and predictive tool for IPF, providing valuable insights for treatment and management.
... For example, a study using the American College of Cardiology Chest Pain-MI registry that used an ML model to predict death after AMI reported an area under the curve (AUC) value of close to 0.9 for each ML model, with extreme gradient boosting (XG-Boost) provide better risk solutions for high-risk individuals [15]. Another ML-based study of adverse event prediction in acute coronary syndrome (apolipoprotein A1/B, ApoA1/B) showed that different machine learning models showed good predictive performance in predicting all-cause death, myocardial infarction, and major bleeding in acute coronary syndrome (ACS) patients at 1 year after discharge, and compared with traditional risk prediction tools, ML algorithm has advantages in predicting MACEs [16]. ...
Background
The study aimed to develop an interpretable machine learning (ML) model to assess and stratify the risk of long-term major adverse cardiovascular events (MACEs) in patients with premature myocardial infarction (PMI) and to analyze the key variables affecting prognosis.
Methods
This prospective study consecutively included patients (male ≤50 years, female ≤55 years) diagnosed with acute myocardial infarction (AMI) at Tianjin Chest Hospital between January 2017 and December 2022. The study endpoint was the occurrence of MACEs during the follow-up period, which was defined as cardiac death, nonfatal stroke, readmission for heart failure, nonfatal recurrent myocardial infarction, and unplanned coronary revascularization. Four machine learning models were built: COX proportional hazards model (COX) regression, random survival forest (RSF), extreme gradient boosting (XGBoost), and DeepSurv. Models were evaluated using concordance index (C-index), Brier score, and decision curve analysis to select the best model for prediction and risk stratification.
Results
A total of 1202 patients with PMI were included, with a median follow-up of 26 months, and MACEs occurred in 200 (16.6%) patients. The RSF model demonstrated the best predictive performance (C-index, 0.815; Brier, 0.125) and could effectively discriminate between high- and low-risk patients. The Kaplan-Meier curve demonstrated that patients categorized as low risk showed a better prognosis (p < 0.0001).
Conclusions
The prognostic model constructed based on RSF can accurately assess and stratify the risk of long-term MACEs in PMI patients. This can help clinicians make more targeted decisions and treatments, thus delaying and reducing the occurrence of poor prognoses.
... However, XG-Boost and meta-classifier models (rather than ANN) were able to better discriminate risk in high-risk populations compared with logistic regression [28]. The use of the ML-based Prediction of Adverse Events Following Acute Coronary Syndrome (PRAISE) score to predict all-cause mortality, MI, and hemorrhage after acute coronary syndrome demonstrated accurate discriminatory ability that can aid clinical decision-making [29]. However, in a real Asian population undergoing percutaneous coronary intervention for acute coronary syndrome, the PRAISE score showed limited potential, with C-index for death, MI, and major hemorrhage of 0.75, 0.61, and 0.62, respectively. ...
Background
Accurate mortality risk prediction is crucial for effective cardiovascular risk management. Recent advancements in artificial intelligence (AI) have demonstrated potential in this specific medical field. Qwen-2 and Llama-3 are high-performance, open-source large language models (LLMs) available online. An artificial neural network (ANN) algorithm derived from the SWEDEHEART (Swedish Web System for Enhancement and Development of Evidence-Based Care in Heart Disease Evaluated According to Recommended Therapies) registry, termed SWEDEHEART-AI, can predict patient prognosis following acute myocardial infarction (AMI).
Objective
This study aims to evaluate the 3 models mentioned above in predicting 1-year all-cause mortality in critically ill patients with AMI.
Methods
The Medical Information Mart for Intensive Care IV (MIMIC-IV) database is a publicly available data set in critical care medicine. We included 2758 patients who were first admitted for AMI and discharged alive. SWEDEHEART-AI calculated the mortality rate based on each patient’s 21 clinical variables. Qwen-2 and Llama-3 analyzed the content of patients’ discharge records and directly provided a 1-decimal value between 0 and 1 to represent 1-year death risk probabilities. The patients’ actual mortality was verified using follow-up data. The predictive performance of the 3 models was assessed and compared using the Harrell C-statistic (C-index), the area under the receiver operating characteristic curve (AUROC), calibration plots, Kaplan-Meier curves, and decision curve analysis.
Results
SWEDEHEART-AI demonstrated strong discrimination in predicting 1-year all-cause mortality in patients with AMI, with a higher C-index than Qwen-2 and Llama-3 (C-index 0.72, 95% CI 0.69-0.74 vs C-index 0.65, 0.62-0.67 vs C-index 0.56, 95% CI 0.53-0.58, respectively; all P<.001 for both comparisons). SWEDEHEART-AI also showed high and consistent AUROC in the time-dependent ROC curve. The death rates calculated by SWEDEHEART-AI were positively correlated with actual mortality, and the 3 risk classes derived from this model showed clear differentiation in the Kaplan-Meier curve (P<.001). Calibration plots indicated that SWEDEHEART-AI tended to overestimate mortality risk, with an observed-to-expected ratio of 0.478. Compared with the LLMs, SWEDEHEART-AI demonstrated positive and greater net benefits at risk thresholds below 19%.
Conclusions
SWEDEHEART-AI, a trained ANN model, demonstrated the best performance, with strong discrimination and clinical utility in predicting 1-year all-cause mortality in patients with AMI from an intensive care cohort. Among the LLMs, Qwen-2 outperformed Llama-3 and showed moderate predictive value. Qwen-2 and SWEDEHEART-AI exhibited comparable classification effectiveness. The future integration of LLMs into clinical decision support systems holds promise for accurate risk stratification in patients with AMI; however, further research is needed to optimize LLM performance and address calibration issues across diverse patient populations.
... D'Ascenzo et al. [7] created and validated a machine learning model, PRAISE, able to predict the risk of all-cause mortality, recurrent myocardial infarction and major bleeding to 1 year after discharge of patients presenting with acute coronary syndrome (ACS). Qaiser et al. [8] proposed a weakly supervised survival convolutional neural network (WSS-CNN) with a visual attention mechanism, to predict cancer patient overall survival from hematoxylin-eosin (H&E)-stained full-slice images. The method does not ask for region level annotation and only relies on patient level survival data for training, making the data preparation much simpler. ...
With the in-depth application of generative AI in clinical statistical processes, the TFL (Table-Figure-Listing) automation platform and macro system have significantly shortened the reporting cycle and improved data quality with unsupervised anomaly detection, laying a clean data foundation for adverse event risk modeling for time-event prediction. On the basis that AI-driven TFL automation and outlier cleaning have significantly improved data quality, we propose Segmented Relative Positional Encoding-Transformer Survival Network (SRPE-TSN): this method only introduces the key improvement of "segmented relative time embedding" on existing Transformer survival models such as SurvTRACE. The longitudinal event sequence of patients was divided into learnable time periods according to clinical milestones, and the relative position information was used to guide multi-head attention to focus on risk signals at different time scales, so as to take into account both right-censored processing and long-term dependency capture. SRPE-TSN increased the 12-month adverse event time-dependent AUC from 0.71 to 0.80 on data from four phase III oncology and cardiovascular trials.
CCS CONCEPTS Applied computing ~ Life and medical sciences ~ Health care information systems
... However, a significant part of the FL literature has been focused on problems related to the federation of neural network-based models [8]- [10], [18], [19]. Although, broadly speaking, neural networks are just the hottest topic in machine learning, there exists a nonnegligible portion of application scenarios in which "classical" machine learning approaches are still used [20] especially when explainability plays an important role [21]- [23]. ...
In recent years, Federated Learning applied to neural networks has garnered significant attention, yet applying this approach to other machine learning algorithms remains underexplored. Support Vector Machines (SVMs), in particular, have seen limited exploration within the federated context, with existing techniques often constrained by the necessity to share the weight vector of the linear classifier. Unfortunately, this constraint severely limits the method’s utility, restricting its application to linear feature spaces. This study addresses and overcomes this limitation by proposing an innovative approach: instead of sharing weight vectors, we advocate sharing support vectors while safeguarding client data privacy through vector perturbation. Simple random perturbation works remarkably well in practice, and indeed we provide a bound on the approximation error of the learnt model which goes to zero as the number of input features grows. We also introduce a refined technique that involves strategically moving the support vectors along the margin of the decision function, which we empirically show to slightly improve the performances. Through extensive experimentation, we demonstrate that our proposed approach achieves state-of-the-art performance and consistently enables the federated classifier to match the performance of classifiers trained on the entire dataset.
... Numerous mortality prediction models target specific diseases [2] or are designed for specific medical settings [3]. However, other models that are not tailored to specific diseases incorporate a multitude of variables that are primarily disease-focused [4]. ...
Background
A poor prognosis within 1 year of discharge is important when making decisions affecting postoperative geriatric inpatients. Comprehensive geriatric assessment (CGA) plays an important role in guiding holistic assessment-based interventions. However, current prognostic models derived from CGA and clinical data are limited and have unsatisfactory performance. We aimed to develop an accurate 1-year mortality prediction model for patients discharged from the geriatric ward using CGA and clinical data.
Methods
This longitudinal cohort study analysed data from 816 consecutively assessed geriatric patients between January 1, 2018 and December 31, 2019. Models were constructed using Cox proportional hazards regression and their validity was assessed by analysing discrimination, calibration, and decision curves. The robustness of the model was determined using sensitivity analysis. A nomogram was developed to predict the 1-year probability of mortality, and the model was validated using C-statistics, Brier scores, and calibration curves.
Results
During 644 patient-years of follow-up, 57 (11·7%) patients died. Clinical variables included in the final prediction model were activities of daily living, serum albumin level, Charlson Comorbidity Index, FRAIL scale, and Mini-Nutrition Assessment-Short Form scores. A C-statistic value of 0·911, a Brier score of 0·058, and a calibration curve validated the model.
Conclusion
Our risk stratification model can accurately predict prospective mortality risk among patients discharged from the geriatric ward. The functionality of this tool facilitates objective palliative care.
... In a similar work from the same group of authors, four ML models (GBM, RF, soft-voting ensemble classifier, and extra tree) were compared (7832 patients for training, 3357 patients for testing) for early prediction of 2-year MACE (cardiac death, non-cardiac death, MI, PCI, and coronary artery bypass grafting) in patients suffering with ACS (long-term prediction); the soft-voting ensemble classifier (SVE) was found to be significantly superior to the other machine learning models, with a calculated AUC of 0.99 [47]. D'Ascenzo et al. developed a machine learning-based model (Adaptive Boosting) using 19,826 patients (training set), known as the PRAISE score, which showed on a cohort of 3444 patients accurate discriminatory capacity in predicting 1-year all-cause mortality (long-term prediction), recurrent myocardial infarction, and major bleeding (defined as Bleeding Academic Research Consortium type 3 or 5) in patients discharged after ACS hospitalization [48]. Moreover, in a population-based study, data from the nationwide SWEDEHEART registry was used to train (11,1558 patients) and test (27,730 patients) an artificial neural network algorithm to predict, in 30,971 patients with ACS from the Western Denmark Heart Registry, 1-year all-cause mortality and 1-year heart failure hospitalizations (long-term predictions). ...
Acute coronary syndrome (ACS) is a global health concern that requires rapid and accurate diagnosis for timely intervention and better patient outcomes. With the emergence of Artificial Intelligence (AI), significant advancements have been made in improving diagnostic accuracy, efficiency, and risk stratification in ACS management. This narrative review examines the current landscape of AI applications in ACS diagnosis and risk stratification, emphasizing key methodologies, technical and clinical implementation challenges, and also possible future research directions. Moreover, unlike previous reviews, this paper also focuses on ethical and legal issues and the feasibility of clinical applications.
... These studies highlight the significant potential of deep learning in disease detection, diagnostics, and clinical decision support. Traditional methods that process IMTS into regular data for standard machine learning applications often fall short due to their inability to handle irregular time intervals effectively [22][23][24][25][26]. The natural and common occurrence of irregular sampling in EHRs, driven by patient status and symptom changes, underscores the importance of developing a model capable of handling these irregularities. ...
Abnormal serum sodium levels are a common and severe complication in stroke patients, significantly increasing mortality risk and prolonging ICU stays. Accurate real-time prediction of serum sodium fluctuations is crucial for optimizing clinical interventions. However, existing predictive models face limitations in handling complex dynamic features and long time series data, making them less effective in guiding individualized treatment. To address this challenge, this study developed a deep learning model based on a multi-head attention mechanism to enable real-time prediction of serum sodium concentrations and provide personalized intervention recommendations for ICU stroke patients. This study utilized publicly available MIMIC-III (n = 2346) and MIMIC-IV (n = 896) datasets, extracting time series data from 10 key clinical indicators closely associated with serum sodium levels. To address the complexity of long time series data, a moving sliding window sub-sampling segmentation method was employed, effectively transforming extensive sequences into more manageable inputs while preserving critical temporal dependencies. By leveraging advanced mathematical modeling, meaningful insights were extracted from sparse and irregular time series data. The resulting time-feature fusion multi-head attention (TFF-MHA) model underwent rigorous validation using public datasets and demonstrated superior performance in predicting both serum sodium values and corresponding intervention measures compared to existing models. This study contributes to the field of healthcare informatics by introducing an innovative, data-driven approach for dynamic serum sodium prediction and intervention recommendation, providing a valuable clinical decision-support tool for optimizing sodium management strategies in critically ill stroke patients.
... In recent years, the rapid development of machine learning technology and its application in clinical data analysis have gained significant attention from medical institutions and researchers for their accurate and efficient predictive performance and clinical decision-making. By comprehensively analyzing large volumes of clinical and biochemical data, machine learning technology identifies risk factors that may be overlooked by traditional methods and plays a crucial role in guiding healthcare professionals to conduct more accurate risk assessments and make informed clinical decisions [8]. According to the results of previous studies, machine learning models have unlimited potential and value in dealing with complex and high-dimensional cardiovascular disease data and postoperative complication risk prediction [9]. ...
Background
This study aimed to develop and validate a predictive model for major adverse cardiovascular events (MACE) following percutaneous coronary intervention (PCI) in patients with new-onset ST-segment elevation myocardial infarction (STEMI) using four machine learning (ML) algorithms.
Methods
Data from 250 new-onset STEMI patients were retrospectively collected. Feature selection was performed using the Boruta algorithm. Four ML algorithms—K-nearest neighbors (KNN), support vector machine (SVM), Complement Naive Bayes (CNB), and logistic regression—were applied to predict MACE risk. Model performance was evaluated using area under the curve (AUC), sensitivity, and specificity. Shapley Additive Explanations (SHAP) analysis was used to rank feature importance, and a nomogram was constructed for risk visualization.
Results
Logistic regression showed the best performance (AUC = 0.814 in training, 0.776 in validation) compared to KNN, SVM, and CNB. SHAP analysis identified seven key predictors, including Killip classification, Gensini score, blood urea nitrogen (BUN), heart rate (HR), creatinine (CR), glutamine transferase (GLT), and platelet count (PCT). The nomogram provided accurate risk predictions with strong agreement between predicted and observed outcomes.
Conclusions
The logistic regression model effectively predicts MACE risk after PCI in STEMI patients. The nomogram serves as a practical tool for clinicians, supporting personalized risk assessment and improving clinical decision-making.
... 69 risk factors including demographics, prior medical history, prior medication, clinical presentation, basic laboratory data with admission were selected and evaluated. D'Ascenzo et al. [31] developed the PRAISE risk scores, a machine learning tool validated with external cohorts, which showed high accuracy in predicting postdischarge outcomes for ACS patients. The 25 risk factors included 16 clinical variables, 5 therapeutic variables, and 2 angiographic variables. ...
Acute Coronary Syndromes (ACS), including ST-segment elevation myocardial infarctions (STEMI) and non-ST-segment elevation myocardial infarctions (NSTEMI), remain a leading cause of mortality worldwide. Traditional cardiovascular risk scores rely primarily on clinical data, often overlooking environmental influences like air pollution that significantly impact heart health. Moreover, integrating complex time-series environmental data with clinical records is challenging. We introduce TabulaTime, a multimodal deep learning framework that enhances ACS risk prediction by combining clinical risk factors with air pollution data. TabulaTime features three key innovations: First, it integrates time-series air pollution data with clinical tabular data to improve prediction accuracy. Second, its PatchRWKV module automatically extracts complex temporal patterns, overcoming limitations of traditional feature engineering while maintaining linear computational complexity. Third, attention mechanisms enhance interpretability by revealing interactions between clinical and environmental factors. Experimental results show that TabulaTime improves prediction accuracy by over 20% compared to conventional models such as CatBoost, Random Forest, and LightGBM, with air pollution data alone contributing over a 10% improvement. Feature importance analysis identifies critical predictors including previous angina, systolic blood pressure, PM10, and NO2. Overall, TabulaTime bridges clinical and environmental insights, supporting personalized prevention strategies and informing public health policies to mitigate ACS risk.
... We fully agree that including additional predictors, such as hyperuricemia, chronic obstructive pulmonary disease, and specific electrocardiographic/echocardiographic parameters may potentially improve the ability of detecting atrial fibrillation (AF) and ventricular arrhythmias (VA) [1]. However, in our study we focused on the application of the PRAISE (PRedicting with Artificial Intelligence riSk aftEr acute coronary syndrome) score model [2] to ensure consistency with the original validation and clinical interpretability of the results. This approach allowed us to evaluate the performance of the model in a real-world context without introducing additional variables that, if incorporated in the score, would have required its recalibration and revalidation. ...
... short-and long-term mortality after an MI due to its association with comorbidities and frailty [2]. ...
... It is an important branch of the field of artificial intelligence and has been widely used in disease prediction [8][9][10]. ML can effectively handle complex high-dimensional data, automatically identify nonlinear relationships and interactions between variables, and significantly improve the accuracy of disease risk prediction [11]. Compared with traditional methods, ML has demonstrated greater accuracy and efficiency in CVD prediction, which is especially suitable for multifactorial and multidimensional disease risk assessments [12,13]. ...
Background
Due to the ageing population and evolving lifestyles occurring in China, middle-aged and elderly populations have become high-risk groups for cardiovascular disease (CVD). The aim of this study was to analyse the incidence characteristics of CVD in these populations and develop a prediction model by using data from the China Health and Retirement Longitudinal Study (CHARLS).
Methods
We used follow-up data from the CHARLS to analyse CVD incidence in the Chinese middle-aged and elderly population over a time span of 9 years. Five machine learning (ML) algorithms were employed for risk prediction. Data preprocessing included missing value imputation via random forest. Feature selection was performed using the Least Absolute Shrinkage and Selection Operator (Lasso CV) method with cross-validation prior to model training. The application of the synthetic minority over-sampling technique (SMOTE) to address class imbalance. Model performance was evaluated via analyses including the area under the ROC curve (AUC), precision, recall, F1 score, and SHAP plots for interpretability.
Results
In accordance with the exclusion criteria, 12,580, 12,061, 11,545, and 11,619 participants were enrolled in four follow-up rounds. The cumulative incidence (CI) of CVD at 2, 4, 7, and 9 years was 2.846%, 8.971%, 17.869% and 20.518%,, respectively. Significant differences in CVD incidence were observed across gender, age, ethnicity, and region, with higher rates observed in females and in the northeast region. Ultimately, 8,080 participants and 24 features were analysed for CVD risk prediction. Five ML models were built based on these features. Although the LGB model achieves an AUC of 0.818, indicating strong overall performance, its F1 score and recall rate are relatively low, at 0.509 and 43.1%, respectively. Shapley additive explanations (SHAP) analyses revealed the importance of key features, such as night sleep duration, TG levels, and waist circumference, in predicting outcomes, and highlighted the nonlinear relationships between these features and CVD risk.
Conclusions
Gender, age, ethnicity, and region are significant factors influencing CVD incidence. Although the LGB model demonstrates good overall performance, its low F1 score and recall rate reveal limitations in identifying high-risk cardiovascular disease patients.
... The exclusion criteria were: (1) postoperative pathology confirming non-gastric primary tumors, (2) distant metastasis, (3) incomplete clinical data, and (4) the detection of other concurrent malignancies within five years. We employed a random sampling strategy to allocate patients from one of the clinical centers into training and internal validation sets at an 8:2 ratio (n = 1,367) [18,19]. In particular, the 'sample' function was used to randomly select 80% of the samples as the training set for the development and enhancement of three machine learning models. ...
Background
The potential of the application of artificial intelligence and biochemical markers of oxidative stress to predict the prognosis of older patients with gastric cancer (GC) remains unclear.
Methods
This retrospective multicenter study included consecutive patients with GC aged ≥ 65 years treated between January 2012 and April 2018. The patients were allocated into three cohorts (training, internal, and external validation). The GC-Integrated Oxidative Stress Score (GIOSS) was developed using Cox regression to correlate biochemical markers with patient prognosis. Predictive models for five-year overall survival (OS) were constructed using random forest (RF), decision tree (DT), and support vector machine (SVM) methods, and validated using area under the curve (AUC) and calibration plots. The SHapley Additive exPlanations (SHAP) method was used for model interpretation.
Results
This study included a total of 1,859 older patients. The results demonstrated that a low GIOSS was a predictor of poor prognosis. RF was the most efficient method, with AUCs of 0.999, 0.869, and 0.796 in the training, internal validation, and external validation sets, respectively. The DT and SVM models showed low AUC values. Calibration and decision curve analyses demonstrated the considerable clinical usefulness of the RF model. The SHAP results identified pN, pT, perineural invasion, tumor size, and GIOSS as key predictive features. An online web calculator was constructed based on the best model.
Conclusions
Incorporating the GIOSS, the RF model effectively predicts postoperative OS in older patients with GC and is a robust prognostic tool. Our findings emphasize the importance of oxidative stress in cancer prognosis and provide a pathway for improved management of GC.
Trial registration
Retrospectively registered at ClinicalTrials.gov (trial registration number: NCT06208046, date of registration: 2024–05-01).
... Machine learning (ML), as a subset of artificial intelligence (AI), enables computers to learn from data, identify patterns, and make predictions or take actions based on acquired knowledge. ML techniques manifest a potential solution to address the limitations of current analytical methods, which can effectively handle multidimensional variables, identify non-linear relationships between features and outcomes, and develop prediction models with improved accuracy and efficiency [13,14]. In 2021, Abuhelwa et al. devised five ML models for prognosticating the survival outcomes of urothelial cancer patients treated with atezolizumab, an immune checkpoint inhibitor. ...
Background
Esophageal squamous cell carcinoma (ESCC) is a highly aggressive malignancy, and current postoperative prognostic assessment methods remain unsatisfactory, underlining the urgent to develop a reliable approach for precision medicine. Given the similarities with gametogenesis, cancer/testis genes (CTGs) are acknowledged for regulation unrestrained multiplication and immune microenvironment during oncogenic processes. These processes are associated with advanced disease and poorer prognosis, indicating that CTGs could serve as ideal prognostic biomarkers in ESCC. The purpose of this study is to develop a novel clinically prognostic prediction system to facilitate the individualized postoperative care.
Methods
We conducted LASSO regression analysis of protein-coding CTGs and clinical characteristics from 119 pathologically confirmed ESCC patients to recognize powerful predictive variables. We employed nine supervised machine learning classifiers and integrated best predictive machine learning classifiers by weighted voting method to construct an ensemble model called PPMESCC. Additionally, functional assay was conducted to examine the potential effect of top-ranking CTG HENMT1 in ESCC.
Results
LASSO regression identified five CTGs and TNM stage as optimized prognostic features. Six machine learning classifiers were integrated to construct an ensemble model, PPMESCC, which exhibited outstanding performance in ESCC prediction. The AUC for PPMESCC was 0.9828 (95% confidence interval: 0.9608 to 0.9926), with an accuracy of 98.32% (95% CI: 96.64–99.16%) in the discovery cohort and 0.9057 (95% CI: 0.8897 to 0.9583) of AUC with an accuracy of 90% (95% CI: 89.08–93.28%) in validation cohort. In addition, the top-ranking CTG HENMT1 encodes 2’-O-methyltransferase of piRNAs that was confirmed positively correlated with the proliferation capacity of ESCC cells. Then we systematically screen piRNAs associated with esophageal carcinoma based on GWAS, eQTL-piRNA, and i2OM databases, and successfully discovered 8 piRNAs potentially regulated by HENMT1.
Conclusion
The study highlights the clinical utility of PPMESCC algorithm in prognostic prediction that may facilitate to establish the personalized screening and management strategies for postoperative ESCC patients.
... A recent study utilised a machine learning intervention to accurately identify patients with acute coronary syndrome. 26 This machine learning approach was feasible and effective and may be useful in guiding clinical decision-making during transitions of care. ...
... As the authors state in their manuscript: '…more sophisticated prediction models such as those involving machine learning (ML), may have provided improved model performance…' 15 The use of ML algorithms on large datasets has shown encouraging results in disease prediction. [17][18][19][20][21][22][23][24] To the best of our knowledge, there is no externally validated scoring system using ML to predict the risk of CS in ACS available to date. The aim of our study was to derive and externally validate a simple scoring system utilizing ML based on variables readily available at first medical contact to identify the risk of developing CS during hospitalization in patients admitted due to ACS. ...
Aims
Cardiogenic shock (CS) is a severe complication of acute coronary syndrome (ACS) with mortality rates approaching 50%. The ability to identify high-risk patients prior to the development of CS may allow for pre-emptive measures to prevent the development of CS. The objective was to derive and externally validate a simple, machine learning (ML)-based scoring system using variables readily available at first medical contact to predict the risk of developing CS during hospitalization in patients with ACS.
Methods and results
Observational multicentre study on ACS patients hospitalized at intensive care units. Derivation cohort included over 40 000 patients from Beth Israel Deaconess Medical Center, Boston, USA. Validation cohort included 5123 patients from the Sheba Medical Center, Ramat Gan, Israel. The final derivation cohort consisted of 3228 and the final validation cohort of 4904 ACS patients without CS at hospital admission. Development of CS was adjudicated manually based on the patients’ reports. From nine ML models based on 13 variables (heart rate, respiratory rate, oxygen saturation, blood glucose level, systolic blood pressure, age, sex, shock index, heart rhythm, type of ACS, history of hypertension, congestive heart failure, and hypercholesterolaemia), logistic regression with elastic net regularization had the highest externally validated predictive performance (c-statistics: 0.844, 95% CI, 0.841–0.847).
Conclusion
STOP SHOCK score is a simple ML-based tool available at first medical contact showing high performance for prediction of developing CS during hospitalization in ACS patients. The web application is available at https://stopshock.org/#calculator.
... Furthermore, applying computer algorithms and ML methods to large data sets with system-level factors may overcome limitations of standard analytic approaches and better capture risk with highly variable features and nonlinear relationships. 10,11 Boston Children's Hospital (BCH) and the MITRE Corporation collaborated to develop supervised ML models predicting the risk for patient harm in the CCC laboratory. Applying advanced analytics to enterprise-level data sets enhances our understanding of patient, procedural, and system-level risk factors and serves as a modern approach to risk assessment in this patient population. ...
Background
Traditional statistical methodologies inadequately capture the complexities of real-world practice to assess risk in congenital cardiac catheterization (CCC). Artificial intelligence and machine learning (ML) techniques are well-suited to analyze preprocedural patient risk given the complexity and heterogeneity of infrequently performed CCC procedures. We sought to apply supervised ML analytics to an enterprise-level data set to enhance understanding of patient-, procedural-, and system-level risk in patients undergoing CCC.
Methods
A comprehensive data set built from electronic health record metadata captured important patient-, procedural-, and system-level characteristics from 2019 through 2020 at Boston Children's Hospital for all patients undergoing diagnostic-only or interventional CCC. Supervised ML was used to develop random forest and least absolute shrinkage and selection operator (LASSO) models to predict the outcome of clinically meaningful adverse events. Models were trained on a randomly selected portion of the data set (75%) whereas the remaining data set (25%) was used for testing purposes. Model performance was evaluated using area under receiver operating characteristic curve and a plot showing the calibration between predicted probability deciles and observed probabilities of the model. Feature importance was assessed.
Results
Our analysis included 1424 cases. Area under the receiver operating characteristic curve for the random forest and LASSO models were 0.67 and 0.68, respectively. Both algorithms exhibited better than random predictive ability with the LASSO model showing a superior level of calibration.
Conclusions
Improving our understanding of risk during preprocedural assessment will inform clinical decision-making and allow for implementation of targeted risk mitigation strategies in high-risk patients to improve CCC patient outcomes.
... ML can be used to directly compare the accuracy of two or more quantitative tests for the same disease/condition [21], playing a role in formulating diagnosis and treatment rules [22][23][24]. ML algorithms have also been used to construct risk forecast models that predict the hazard ratio of adverse events [25,26] or predict the classification of double-class/multiclass endpoints at a specific time [27]. Nevertheless, the time interval of occurrence of specific oncological outcomes for CRC patients cannot be vertically predicted in these models, and some of the models' important variables are unknown, casting doubt on their clinical credibility. ...
BACKGROUND
Colorectal cancer (CRC) is characterized by high heterogeneity, aggressiveness, and high morbidity and mortality rates. With machine learning (ML) algorithms, patient, tumor, and treatment features can be used to develop and validate models for predicting survival. In addition, important variables can be screened and different applications can be provided that could serve as vital references when making clinical decisions and potentially improving patient outcomes in clinical settings.
AIM
To construct prognostic prediction models and screen important variables for patients with stage I to III CRC.
METHODS
More than 1000 postoperative CRC patients were grouped according to survival time (with cutoff values of 3 years and 5 years) and assigned to training and testing cohorts (7:3). For each 3-category survival time, predictions were made by 4 ML algorithms (all-variable and important variable-only datasets), each of which was validated via 5-fold cross-validation and bootstrap validation. Important variables were screened with multivariable regression methods. Model performance was evaluated and compared before and after variable screening with the area under the curve (AUC). SHapley Additive exPlanations (SHAP) further demonstrated the impact of important variables on model decision-making. Nomograms were constructed for practical model application.
RESULTS
Our ML models performed well; the model performance before and after important parameter identification was consistent, and variable screening was effective. The highest pre- and postscreening model AUCs 95% confidence intervals in the testing set were 0.87 (0.81-0.92) and 0.89 (0.84-0.93) for overall survival, 0.75 (0.69-0.82) and 0.73 (0.64-0.81) for disease-free survival, 0.95 (0.88-1.00) and 0.88 (0.75-0.97) for recurrence-free survival, and 0.76 (0.47-0.95) and 0.80 (0.53-0.94) for distant metastasis-free survival. Repeated cross-validation and bootstrap validation were performed in both the training and testing datasets. The SHAP values of the important variables were consistent with the clinicopathological characteristics of patients with tumors. The nomograms were created.
CONCLUSION
We constructed a comprehensive, high-accuracy, important variable-based ML architecture for predicting the 3-category survival times. This architecture could serve as a vital reference for managing CRC patients.
... In recent years, several scoring systems have been employed to assess bleeding in ACS patients. The PARIS, PRECISE DAPT, and PRAISE risk scores are primarily used to evaluate the risk of outpatient bleeding in patients with ACS [8][9][10]. Although the CRUSADE bleeding risk score is the most extensively employed in-hospital bleeding scoring tool in clinical settings [11,12], it does not account for patients using anticoagulants. ...
Background
The Academic Research Consortium for High Bleeding Risk (ARC-HBR) criteria were proposed for predicting bleeding risk in patients undergoing percutaneous coronary intervention (PCI). However, there is a lack of research evaluating the risk of in-hospital bleeding following PCI for acute coronary syndrome (ACS) utilizing the ARC-HBR criteria.
Methods and results
This study involved 1013 ACS patients who underwent PCI and dual antiplatelet therapy. There were 63 cases of in-hospital bleeding events (6.22 %). According to the ARC-HBR criteria, patients classified as HBR had a significantly greater bleeding rate than non-HBR patients (15.81 % vs. 1.99 %, p < 0.001). As the CRUSADE score category increased, the risk of bleeding also increased. The area under the receiver operating characteristic curve (AUC) of the ARC-HBR criteria was significantly greater than that of the CRUSADE score for bleeding (0.751 vs. 0.696, p < 0.0001). Subgroup analysis revealed that the ARC-HBR criteria exhibited better predictive ability for ST-segment elevation myocardial infarction (STEMI, AUC 0.767 vs. 0.694, p = 0.020) but comparable predictive ability in patients with unstable angina (AUC 0.756 vs. 0.644, p = 0.213), non-ST-segment elevation myocardial infarction (AUC 0.713 vs. 0.683, p = 0.644), and non-ST-segment elevation ACS (AUC 0.739 vs. 0.687, p = 0.330).
Conclusion
Compared with the CRUSADE score, the ARC-HBR criteria demonstrate superior predictive ability for in-hospital bleeding events during PCI in ACS patients. Routine assessment of the ARC-HBR score might be helpful for identifying high-risk individuals in this specific population.
Despite advances in research and patient management, atherosclerosis and its dreaded acute and chronic sequelae continue to account for one out of three deaths globally. The vast majority of acute coronary syndromes (ACS) arise from either plaque rupture or erosion, but other mechanisms, including calcific nodules, embolism, spontaneous coronary artery dissection, coronary spasm, and microvascular dysfunction, can also cause ACS. This ACS heterogeneity necessitates a paradigm shift in its management that extends beyond the binary interpretation of electrocardiographic and biomarker data. Indeed, given the evolution in the global risk factor profile, the increasing importance of previously underappreciated mechanisms, the evolving appreciation of sex-specific disease characteristics, and the advent of rapidly evolving technologies, a precision medicine approach is warranted. This review provides an update of the mechanisms of ACS, delineates the role of previously underappreciated contributors, discusses sex-specific differences, and explores novel tools for contemporary and personalized management of patients with ACS. Beyond mechanistic insights, it examines evolving imaging techniques, biomarkers, and regression- and machine learning-based approaches for the diagnosis (e.g. CoDE-ACS, MI3) and prognosis (e.g. PRAISE, GRACE, SEX-SHOCK scores) of ACS, along with their implications for future ACS management. A more individualized approach to patients with ACS is advocated, emphasizing the need for innovative studies on emerging technologies, including artificial intelligence, which may collectively facilitate clinical decision-making within a more mechanistic framework, thereby personalizing patient care and potentially improving long-term outcomes.
Artificial intelligence (AI), a branch of computer science focused on developing algorithms that replicate intelligent behaviour, has recently been used in patients management by enhancing diagnostic and prognostic capabilities of various resources such as hospital datasets, electrocardiograms and echocardiographic acquisitions. Machine learning (ML) and deep learning (DL) models, both key subsets of AI, have demonstrated robust applications across several cardiovascular diseases, from the most diffuse like hypertension and ischemic heart disease to the rare infiltrative cardiomyopathies, as well as to estimation of LDL cholesterol which can be achieved with better accuracy through AI.
Additional emerging applications are encountered when unsupervised ML methodology shows promising results in identifying distinct clusters or phenotypes of patients with atrial fibrillation that may have different risks of stroke and response to therapy.
Interestingly, since ML techniques do not analyse the possibility that a specific pathology can occur but rather the trajectory of each subject and the chain of events that lead to the occurrence of various cardiovascular pathologies, it has been considered that DL, by resembling the complexity of human brain and using artificial neural networks, might support clinical management through the processing of large amounts of complex information; however, external validity of algorithms cannot be taken for granted, while interpretability of the results may be an issue, also known as a “black box” problem. Notwithstanding these considerations, facilities and governments are willing to unlock the potential of AI in order to reach the final step of healthcare advancements while ensuring that patient safety and equity are preserved.
Objective
We developed a risk stratification model to predict serious adverse hospitalization events (mortality, cardiac shock, cardiac arrest) (SAHE) after acute coronary syndrome (ACS) based on machine-learning models and logistic regression model.
Methods
This cohort study is based on the CCC-ACS project. The primary efficacy outcomes were SAHE. Clinical prediction models were established based on five machine-learning (XGBoost, RF, MLP, KNN, and stacking model) and logistic regression models.
Results
Among the 112 363 patients in the study, age (55–65 years: OR: 1.392; 95%CI: 1.212–1.600; 65–75 years: OR: 1.878; 95%CI: 1.647–2.144; ≥75 year: OR: 2.976; 95%CI: 2.615–3.393), history of diabetes mellitus (OR: 1.188; 95%CI: 1.083–1.302), history of renal failure (OR: 1.645; 95%CI: 1.311–2.044), heart rate (60–100 beats/min: OR: 0.468; 95%CI: 0.409–0.536; ≥100 beats/min: OR: 0.540; 95%CI: 0.454–0.643), shock index (0.4–0.8: OR: 1.796; 95%CI: 1.440–2.264; ≥0.8: OR: 5.883; 95%CI: 4.619–7.561), KILLIP (II: OR: 1.171; 95%CI: 1.048–1.306; III: OR: 1.696; 95%CI: 1.469–1.952; IV: OR: 7.811; 95%CI: 7.023–8.684), and cardiac arrest at admission (OR: 12.507; 95%CI: 10.757–14.530) were independent predictors of severe adverse hospitalization events for ACS patients. In several machine-learning models, RF (AUC: 0.817; 95%CI: 0.808–0.826) and XGBoost (AUC: 0.816; 95%CI: 0.807–0.825) also showed good discrimination in the training set, which ranked the first two positions. They also presented good accuracy and the best clinical benefits in the decision curve analysis. In addition, logistic regression was able to discriminate the SAHE (AUC: 0.816; 95%CI: 0.807–0.825) and performed the best prediction accuracy (0.822; 95%CI: 0.822–0.822) compared to several machine-learning models. Model calibration and decision curve analysis showed these prediction models have similar predictive performance. Based on these findings, we developed two CCC-ACS In-hospital Major Adverse Events Risk Scores and its online calculator. One is based on machine-learning model (https://ccc-acs-sae-3-xcnjsvoccusjwkfhfthh44.streamlit.app/), and another is based on logistic regression model (https://ccc-acs-sae-logistic-9te57ylnq3kazkeuyc7dub.streamlit.app/), offering a validated tool to predict survival for patients with ACS during hospitalization.
Conclusions
Machine-learning-based approaches for identifying predictors of SAHE after an ACS were feasible and practical. Based on this, we developed two online risk prediction websites for clinicians’ decision-making. The CCC-ACS-MSAE score showed accurate discriminative capabilities for predicting severe adverse hospitalization events and might help guide clinical decision-making.
Key messages: Three research questions and three bullet points What is already known on this topic? Observational studies have identified risk factors for in-hospital death in patients with acute coronary syndromes (ACS). However, the real-world results of a large sample in China still need to be further explored. What does this study add? Machine-learning-based approaches for identifying predictors of SAHE after an ACS were feasible and practical. Based on these findings, we developed two CCC-ACS In-hospital Major Adverse Events Risk Scores and its online calculator. One is based on machine-learning model (https://ccc-acs-sae-3-xcnjsvoccusjwkfhfthh44.streamlit.app/), and another is based on logistic regression model (https://ccc-acs-sae-logistic-9te57ylnq3kazkeuyc7dub.streamlit.app/), offering a validated tool to predict survival for patients with ACS during hospitalization. How this study might affect research, practice, or policy? Early identification of high-risk ACS patients will help reduce in-hospital deaths and improve the prognosis of ACS patients.
Background
The accuracy of current tools for predicting adverse events in older inpatients with possible sarcopenia is still insufficient to develop individualized nutrition-related management strategies. The objectives were to develop a machine learning model based on nutritional assessment for the prediction of all-cause death and infectious complications.
Methods
A cohort of older patients with possible sarcopenia (divided into training group [70%] and validation group [30%]) from 30 hospitals in 14 major cities in China was retrospectively analyzed. Clinical characteristics, laboratory examination, Nutritional risk Screening-2002 (NRS-2002) and mini-nutritional Assessment-Short form (MNA-SF) were used to construct machine learning models to predict in-hospital adverse events, including all-cause mortality and infectious complications. The applied algorithms included decision tree, random forest, gradient boosting machine (GBM), LightGBM, extreme gradient boosting and neural network. Model performance was assessed according to learning a series of learning metrics including area under the receiver operating characteristic curve (AUC) and accuracy.
Results
Among 3 999 participants (mean age 75.89 years [SD 7.14]; 1 805 [45.1%] were female), 373 (9.7%) had adverse events, including 62 (1.6%) of in-hospital death and 330 (8.5%) of infectious complications. The decision tree model showed a better AUC of 0.7072 (95% CI 0.6558–0.7586) in the validation cohort, using the five most important variables (i.e., mobility, reduced food intake, white blood cell count, upper arm circumference, and hypoalbuminemia).
Conclusions
Machine learning prediction models are feasible and effective for identifying adverse events, and may be helpful to guide clinical nutrition decision-making in older inpatients with possible sarcopenia.
This study aimed to investigate the risk factors for low postoperative blood pressure and construct a machine learning (ML) model based on these features for real-time prediction in patients with oral cancer following reconstruction surgery. The retrospective cohort analysis included adults who had undergone oral cancer resection and free flap reconstruction surgery between December 2022 and December 2023. Patient clinical characteristics were obtained from the electronic medical records. Seven ML techniques were attempted with postoperative hypotension (POH) (mean arterial pressure ˂ 55 mmHg) as the primary outcome. The best-performing ML model was tuned, and the final performance was evaluated using split-set validation, followed by risk factor identification and model interpretability. Of the 727 patients, 412 were finally included, with 66 (16.2%) experiencing POH, resulting in higher inpatient costs and prolonged hospitalization. With an area under the receiver operating characteristic curve of 0.805 (95% confidence interval [CI]: 0.674–0.935), the random forest model demonstrated excellent performance. Shapley additive explanation and feature importance analysis revealed that systolic pressure, heart rate, tumor size, lactic acid level, diastolic pressure, surgical time, total liquid infusion volume, and body mass index were significant risk factors for POH, indicating the robustness of the random forest model.
Background
Severe functional mitral regurgitation (FMR) may benefit from mitral Transcathter Edge to Egde Repair (TEER), but selection of patients remains to be optimised.
Objectives
The aim of this study was to use Machine-Learning (ML) approaches to uncover concealed connections between clinical, echocardiographic, and hemodynamic data associated with patients’ outcomes.
Methods
Consecutive patients undergoing TEER from 2009 to 2020 were included in the MITRA-AI registry. The primary endpoint was a composite of cardiovascular death or heart failure hospitalisation at one year. External validation was performed on the Mitrascore cohort.
Results
822 patients were included. The composite primary endpoint occurred in 250 (30%) patients. Four clusters with decreasing risk of the primary endpoint were identified (42%, 37%, 25% and 20% from cluster 1 to cluster 4, respectively). Clusters were combined into a high-risk (clusters 1 and 2) and a low-risk phenotype (clusters 3 and 4). High-risk phenotype patients had larger LVs (> 107 ml/m2), lower LVEF (< 35%) and more prevalent ischemic aetiology compared to low-risk phenotype patients. Within low.risk groups, permanent atrial fibrillation amplified that of HF hospitalizations. In the Mitrascore cohort, the incidence of the primary endpoint was 48%, 52%, 35% and 42% across clusters.
Conclusions
A ML analysis identified meaningful clinical phenotypic presentations in FMR undergoing TEER, with significant differences in terms of cardiovascular death and heart failure hospitalizations, confirmed in an external validation cohort.
Acute coronary syndromes (ACS) continue to pose significant challenges for clinical practitioners, particularly regarding the prediction of mid- to long-term outcomes. This study aims to investigate the impact of in-hospital bleeding (IHB) at one-year follow-up in patients admitted for ACS. Data from 23,270 patients enrolled in the international PRAISE registry and discharged after ACS were analyzed. A total of 1,060 patients experienced IHB, while 18,765 did not; 3,445 were excluded due to missing data. The primary endpoint was all-cause mortality at 1 year. Secondary endpoints included major bleeding, reinfarction, and composite endpoints at 1 year. Patients with IHB were older, more frequently female, and had a higher prevalence of cardiovascular risk factors (all p < 0.05). At discharge, IHB patients were less likely to receive optimal medical therapy. At the one-year follow-up, all-cause mortality, major bleeding, and reinfarction were significantly higher in the IHB group (all p < 0.001). Bivariate analysis showed a strong association between IHB and all the outcomes of interest (all OR > 1; all p < 0.001). These associations remained significant even after adjusting for several covariates, except for reinfarction (OR 1.3; 95% CI 0.9–2.11; p = 0.149). Age, female sex, hypertension, and peripheral artery disease were found to be independent predictors of IHB, while DES implantation, radial access and left ventricular ejection fraction were identified as protective factors. IHB is a hallmark of frailty in ACS patients; therefore, greater attention should be given during follow-up to patients experiencing this condition.
Artificial Intelligence (AI) and Machine Learning (ML) are now used in many research and industrial fields. Among these applications, they are widely employed to support medical professionals, patients, and caregivers in disease management, and they appear well-positioned to revolutionize the healthcare industry. Among the many applications of AI are medical imaging, predictive analytics, personalized treatment plans, and administrative automation. AI/ML can manage and integrate different data resources (clinical measurements and observations, biological data, experimental results, environmental information, and wearable device data) into models for managing human diseases. Incorporating AI/ML into medicine can enhance every stage of patient care, from risk stratification to diagnosis and therapy. Consequently, this chapter delves into how AI can positively affect every step of the diagnostic-therapeutic pathway, focusing on patient risk stratification, diagnosis, and disease treatment. Specifically, for each of these domains, we report the general approach of AI/ML in a real-world context. As shown, AI/ML-based approaches can improve healthcare workers’ knowledge, enable them to spend more time on direct patient care, and reduce fatigue.
Acute coronary syndrome (ACS) is a leading cause of death worldwide. Prompt and accurate diagnosis of acute myocardial infarction (AMI) or ACS is crucial for improved management and prognosis of patients. The rapid growth of machine learning (ML) research has significantly enhanced our understanding of ACS. Most studies have focused on applying ML to detect ACS, predict prognosis, manage treatment, identify risk factors, and discover potential biomarkers, particularly using data from electrocardiograms (ECGs), electronic medical records (EMRs), imaging, and omics as the main data modality. Additionally, integrating ML with smart devices such as wearables, smartphones, and sensor technology enables real-time dynamic assessments, enhancing clinical care for patients with ACS. This review provided an overview of the workflow and key concepts of ML as they relate to ACS. It then provides an overview of current ML algorithms used for ACS diagnosis, prognosis, identification of potential risk biomarkers, and management. Furthermore, we discuss the current challenges faced by ML algorithms in this field and how they might be addressed in the future, especially in the context of medicine.
Trastuzumab (Tra)-induced cardiotoxicity (TIC) is a serious side effect of cancer chemotherapy, which can seriously harm the health of cancer patients. However, there is currently a lack of effective and reliable biomarkers for the early diagnosis of TIC in clinical practice. Therefore, we screened the TIC candidate diagnostic gene solute carrier family 6 member 6 (SLC6A6) by combining multi-machine learning algorithm based on bioinformatics. In addition, cross-validation showed that SLC6A6 had a consistent expression trend in multi-data-sets. To further explore the diagnostic capability of SLC6A6 in TIC, we constructed a nomogram diagnostic model based on SLC6A6 expression level, and receiver operating characteristic (ROC) curve, calibration curve and decision curve analysis proved that SLC6A6 had good diagnostic capability. In order to further verify the TIC expression of SLC6A6 in the real world, we have constructed cell and animal models. Animal experiments showed that left ventricular ejection fraction (LVEF) was significantly decreased (from 65.01 ± 3.30% and 351.32 ± 3.51%, p < 0.0001) after Tra injection, and severe cardiac function was impaired. Similarly, RT-QPCR demonstrated that SLC6A6 was significantly downregulated in Tra-treated cardiomyocytes in vitro and in vivo. Our study suggests that the differential expression of SLC6A6 in vitro and in vivo models is associated with TIC, which may be a candidate diagnostic gene for the early occurrence and development of TIC and a potential therapeutic target.
Background A poor prognosis within 1 year of discharge is important when making decisions affecting postoperative geriatric inpatients. Comprehensive geriatric assessment (CGA) plays an important role in guiding holistic assessment-based interventions. However, current prognostic models derived from CGA and clinical data are limited and have unsatisfactory performance. We aimed to develop an accurate 1-year mortality prediction model for patients discharged from the geriatric ward using CGA and clinical data. Methods This longitudinal cohort study analysed data from 816 consecutively assessed geriatric patients between January 1, 2018 and December 31, 2019. Models were constructed using Cox proportional hazards regression and their validity was assessed by analysing discrimination, correction, and decision curves. The robustness of the model was determined using sensitivity analysis. A nomogram was developed to predict the 1-year probability of mortality, and the model was validated using C-statistics, Brier scores, and calibration curves. Results During 644 patient-years of follow-up, 57 (11·7%) patients died. Clinical variables included in the final prediction model were activities of daily living, serum albumin level, Charlson Comorbidity Index, FRAIL scale, and Mini-Nutrition Assessment-Short Form scores. A C-statistic value of 0·911, a Brier score of 0·058, and a calibration curve validated the model. Conclusion Our risk stratification model can accurately predict prospective mortality risk among patients discharged from the geriatric ward. The functionality of this tool facilitates objective palliative care.
Background: Accurate bleeding risk stratification after percutaneous coronary intervention (PCI) is important for treatment individualization. However, there is still an unmet need for a more precise and standardized identification of high bleeding risk patients. We derived and validated a novel bleeding risk score by augmenting the PRECISE-DAPT score with the Academic Research Consortium for High Bleeding Risk (ARC-HBR) criteria.
Methods: The derivation cohort comprised 29,188 patients undergoing PCI, of whom 1136 (3.9%) had a Bleeding Academic Research Consortium (BARC) 3 or 5 bleeding at 1 year, from four contemporary real-world registries and the XIENCE V USA trial. The PRECISE-DAPT score was refitted with a Fine-Gray model in the derivation cohort and extended with the ARC-HBR criteria. The primary outcome was BARC 3 or 5 bleeding within 1 year. Independent predictors of BARC 3 or 5 bleeding were selected at multivariable analysis (p<0.01). The discrimination of the score was internally assessed with apparent validation and cross-validation. The score was externally validated in 4578 patients from the MASTER DAPT trial and 5970 patients from the STOPDAPT-2 total cohort.
Results: The PRECISE-HBR score (age, estimated glomerular filtration rate, hemoglobin, white-blood-cell count, previous bleeding, oral anticoagulation, and ARC-HBR criteria) showed an area under the curve (AUC) for 1-year BARC 3 or 5 bleeding of 0.73 (95% CI, 0.71–0.74) at apparent validation, 0.72 (95% CI, 0.70–0.73) at cross-validation, 0.74 (95% CI, 0.68–0.80) in the MASTER DAPT, and 0.73 (95% CI, 0.66–0.79) in the STOPDAPT-2, with superior discrimination than the PRECISE-DAPT (cross-validation: Δ AUC, 0.01; p=0.02; MASTER DAPT: Δ AUC, 0.05; p=0.004; STOPDAPT-2: Δ AUC, 0.02; p=0.20) and other risk scores. In the derivation cohort, a cut-off of 23 points identified 11,414 patients (39.1%) with a 1-year BARC 3 or 5 bleeding risk ≥4%. An alternative version of the score, including acute myocardial infarction on admission instead of white-blood-cell count, showed similar predictive ability.
Conclusions: The PRECISE-HBR score is a contemporary, simple 7-item risk score to predict bleeding after PCI, offering a moderate improvement in discrimination over multiple existing scores. Further evaluation is required to assess its impact on clinical practice.
Introduction:
The performance of seven cardiovascular (CV) risk algorithms is evaluated in a multicentric cohort of ankylosing spondylitis (AS) patients. Performance and calibration of traditional CV predictors have been compared with the novel paradigm of machine learning (ML).
Methods:
A retrospective analysis of prospectively collected data from an AS cohort has been performed. The primary outcome was the first CV event. The discriminatory ability of the algorithms was evaluated using the area under the receiver operating characteristic (ROC) curve (AUC), which is like the concordance-statistic (c-statistic). Three ML techniques were considered to calculate the CV risk: support vector machine (SVM), random forest (RF), and k-nearest neighbor (KNN).
Results:
Of 133 AS patients enrolled, 18 had a CV event. c-statistic scores of 0.71, 0.61, 0.66, 0.68, 0.66, 0.72, and 0.67 were found, respectively, for SCORE, CUORE, FRS, QRISK2, QRISK3, RRS, and ASSIGN. AUC values for the ML algorithms were: 0.70 for SVM, 0.73 for RF, and 0.64 for KNN. Feature analysis showed that C-reactive protein (CRP) has the highest importance, while SBP and hypertension treatment have lower importance.
Conclusions:
All of the evaluated CV risk algorithms exhibit a poor discriminative ability, except for RRS and SCORE, which showed a fair performance. For the first time, we demonstrated that AS patients do not show the traditional ones used by CV scores and that the most important variable is CRP. The present study contributes to a deeper understanding of CV risk in AS, allowing the development of innovative CV risk patient-specific models.
Aims:
Our aim was to develop a machine learning (ML)-based risk stratification system to predict 1-, 2-, 3-, 4-, and 5-year all-cause mortality from pre-implant parameters of patients undergoing cardiac resynchronization therapy (CRT).
Methods and results:
Multiple ML models were trained on a retrospective database of 1510 patients undergoing CRT implantation to predict 1- to 5-year all-cause mortality. Thirty-three pre-implant clinical features were selected to train the models. The best performing model [SEMMELWEIS-CRT score (perSonalizEd assessMent of estiMatEd risk of mortaLity With machinE learnIng in patientS undergoing CRT implantation)], along with pre-existing scores (Seattle Heart Failure Model, VALID-CRT, EAARN, ScREEN, and CRT-score), was tested on an independent cohort of 158 patients. There were 805 (53%) deaths in the training cohort and 80 (51%) deaths in the test cohort during the 5-year follow-up period. Among the trained classifiers, random forest demonstrated the best performance. For the prediction of 1-, 2-, 3-, 4-, and 5-year mortality, the areas under the receiver operating characteristic curves of the SEMMELWEIS-CRT score were 0.768 (95% CI: 0.674-0.861; P < 0.001), 0.793 (95% CI: 0.718-0.867; P < 0.001), 0.785 (95% CI: 0.711-0.859; P < 0.001), 0.776 (95% CI: 0.703-0.849; P < 0.001), and 0.803 (95% CI: 0.733-0.872; P < 0.001), respectively. The discriminative ability of our model was superior to other evaluated scores.
Conclusion:
The SEMMELWEIS-CRT score (available at semmelweiscrtscore.com) exhibited good discriminative capabilities for the prediction of all-cause death in CRT patients and outperformed the already existing risk scores. By capturing the non-linear association of predictors, the utilization of ML approaches may facilitate optimal candidate selection and prognostication of patients undergoing CRT implantation.
Background:
Frailty has become a high-priority issue in cardiovascular medicine because of the aging of cardiovascular patients. Simple and reproducible tools to assess frailty in elderly patients are clearly on demand. Their application may help physicians in the selection of invasive and medical treatments and in the timing and modality of the follow-up. The frailty in elderly patients receiving cardiac interventional procedures (FRASER) program is designed with the aim to validate the use of the short physical performance battery (SPPB) as prognostic tools in patients admitted to hospital for acute coronary syndrome (ACS).
Methods:
The FRASER program is a multicenter prospective study involving 4 Italian cardiology units. The FRASER program enrolls only patients aged ≥70 years. The core of the FRASER program includes patients admitted to hospital for ACS. The aims are (1) to describe SPPB distribution before hospital discharge and (2) to investigate the prognostic role of SPPB score. The primary outcome is a composite of 1-year all-cause mortality and hospital readmission for any cause. Ancillary analyses will be focused on different study populations (patients hospitalized for arrhythmias or acute heart failure or symptomatic severe aortic stenosis) and on different tools to assess frailty (multidimensional prognostic index, clinical frailty score, grip strength).
Discussion:
The FRASER program will fill critical gaps in the knowledge regarding the link between frailty, cardiovascular disease, interventional procedures and outcome and will help physicians in the generation of a more personalized risk assessment and in the identification of potential targets for interventions.
P2Y12 antagonist therapy improves outcomes in acute myocardial infarction (MI) patients. Novel agents in this class are now available in the US. We studied the introduction of prasugrel into contemporary MI practice to understand the appropriateness of its use and assess for changes in antiplatelet management practices.
Using ACTION Registry-GWTG (Get-with-the-Guidelines), we evaluated patterns of P2Y12 antagonist use within 24 hours of admission in 100 228 ST elevation myocardial infarction (STEMI) and 158 492 Non-ST elevation myocardial infarction (NSTEMI) patients at 548 hospitals between October 2009 and September 2012. Rates of early P2Y12 antagonist use were approximately 90% among STEMI and 57% among NSTEMI patients. From 2009 to 2012, prasugrel use increased significantly from 3% to 18% (5% to 30% in STEMI; 2% to 10% in NSTEMI; P for trend <0.001 for all). During the same period, we observed a decrease in use of early but not discharge P2Y12 antagonist among NSTEMI patients. Although contraindicated, 3.0% of patients with prior stroke received prasugrel. Prasugrel was used in 1.9% of patients ≥75 years and 4.5% of patients with weight <60 kg. In both STEMI and NSTEMI, prasugrel was most frequently used in patients at the lowest predicted risk for bleeding and mortality. Despite lack of supporting evidence, prasugrel was initiated before cardiac catheterization in 18% of NSTEMI patients.
With prasugrel as an antiplatelet treatment option, contemporary practice shows low uptake of prasugrel and delays in P2Y12 antagonist initiation among NSTEMI patients. We also note concerning evidence of inappropriate use of prasugrel, and inadequate targeting of this more potent therapy to maximize the benefit/risk ratio.
Acute coronary syndromes (ACS) represent a difficult challenge for physicians. Risk scores have become the cornerstone in clinical and interventional decision making.
PubMed was systematically searched for ACS risk score studies. They were divided into ACS studies (evaluating Unstable Angina; UA, Non ST Segment Elevation Myocardial Infarction; NSTEMI, and ST Segment Elevation Myocardial Infarction; STEMI), UA/NSTEMI studies or STEMI studies. The c-statistics of validation studies were pooled when appropriate with random-effect methods. 7 derivation studies with 25,525 ACS patients and 15 validation studies including 257,654 people were formally appraised. Pooled analysis of GRACE scores, both at short (0.82; 0.80-0.89 I.C 95%) and long term follow up (0.84; 0.82-0.87; I.C 95%) showed the best performance, with similar results to Simple Risk Index (SRI) derivation cohorts at short term. For NSTEMI/UA, 18 derivation studies with 56,560 patients and 18 validation cohorts with 56,673 patients were included. Pooled analysis of validations studies showed c-statistics of 0.54 (95% CI = 0.52-0.57) and 0.67 (95% CI = 0.62-0.71) for short and long term TIMI validation studies, and 0.83 (95% CI = 0.79-9.87) and 0.80 (95% CI = 0.74-0.89) for short and long term GRACE studies. For STEMI, 15 studies with 134,557 patients with derivation scores, and 17 validation studies with 187,619 patients showed a pooled c-statistic of 0.77 (95% CI = 0.71-0.83) and 0.77 (95% CI = 0.72-0.85) for TIMI at short and long term, and a pooled c-statistic of 0.82 (95% CI = 0.81-0.83) and 0.81 (95% CI = 0.80-0.82) for GRACE at short and long terms respectively.
TIMI and GRACE are the risk scores that up until now have been most extensively investigated, with GRACE performing better. There are other potentially useful ACS risk scores available however these have not undergone rigorous validation. This study suggests that these other scores may be potentially useful and should be further researched.
Treatments for non-ST-segment-elevation myocardial infarction (NSTEMI) reduce ischemic events but increase bleeding. Baseline prediction of bleeding risk can complement ischemic risk prediction for optimization of NSTEMI care; however, existing models are not well suited for this purpose.
We developed (n=71 277) and validated (n=17 857) a model that identifies 8 independent baseline predictors of in-hospital major bleeding among community-treated NSTEMI patients enrolled in the Can Rapid risk stratification of Unstable angina patients Suppress ADverse outcomes with Early implementation of the ACC/AHA guidelines (CRUSADE) Quality Improvement Initiative. Model performance was tested by c statistics in the derivation and validation cohorts and according to postadmission treatment (ie, invasive and antithrombotic therapy). The CRUSADE bleeding score (range 1 to 100 points) was created by assignment of weighted integers that corresponded to the coefficient of each variable. The rate of major bleeding increased by bleeding risk score quintiles: 3.1% for those at very low risk (score < or = 20); 5.5% for those at low risk (score 21-30); 8.6% for those at moderate risk (score 31-40); 11.9% for those at high risk (score 41-50); and 19.5% for those at very high risk (score >50; P(trend) <0.001). The c statistics for the major bleeding model (derivation=0.72 and validation=0.71) and risk score (derivation=0.71 and validation=0.70) were similar. The c statistics for the model among treatment subgroups were as follows: > or = 2 antithrombotics=0.72; <2 antithrombotics=0.73; invasive approach=0.73; conservative approach=0.68.
The CRUSADE bleeding score quantifies risk for in-hospital major bleeding across all postadmission treatments, which enhances baseline risk assessment for NSTEMI care.
Accurate estimation of risk for untoward outcomes after patients have been hospitalized for an acute coronary syndrome (ACS) may help clinicians guide the type and intensity of therapy.
To develop a simple decision tool for bedside risk estimation of 6-month mortality in patients surviving admission for an ACS.
A multinational registry, involving 94 hospitals in 14 countries, that used data from the Global Registry of Acute Coronary Events (GRACE) to develop and validate a multivariable stepwise regression model for death during 6 months postdischarge. From 17,142 patients presenting with an ACS from April 1, 1999, to March 31, 2002, and discharged alive, 15,007 (87.5%) had complete 6-month follow-up and represented the development cohort for a model that was subsequently tested on a validation cohort of 7638 patients admitted from April 1, 2002, to December 31, 2003.
All-cause mortality during 6 months postdischarge after admission for an ACS.
The 6-month mortality rates were similar in the development (n = 717; 4.8%) and validation cohorts (n = 331; 4.7%). The risk-prediction tool for all forms of ACS identified 9 variables predictive of 6-month mortality: older age, history of myocardial infarction, history of heart failure, increased pulse rate at presentation, lower systolic blood pressure at presentation, elevated initial serum creatinine level, elevated initial serum cardiac biomarker levels, ST-segment depression on presenting electrocardiogram, and not having a percutaneous coronary intervention performed in hospital. The c statistics for the development and validation cohorts were 0.81 and 0.75, respectively.
The GRACE 6-month postdischarge prediction model is a simple, robust tool for predicting mortality in patients with ACS. Clinicians may find it simple to use and applicable to clinical practice.
Concurrent advances in information technology infrastructure and mobile computing power in many low and middle-income countries (LMICs) have raised hopes that artificial intelligence (AI) might help to address challenges unique to the field of global health and accelerate achievement of the health-related sustainable development goals. A series of fundamental questions have been raised about AI-driven health interventions, and whether the tools, methods, and protections traditionally used to make ethical and evidence-based decisions about new technologies can be applied to AI. Deployment of AI has already begun for a broad range of health issues common to LMICs, with interventions focused primarily on communicable diseases, including tuberculosis and malaria. Types of AI vary, but most use some form of machine learning or signal processing. Several types of machine learning methods are frequently used together, as is machine learning with other approaches, most often signal processing. AI-driven health interventions fit into four categories relevant to global health researchers: (1) diagnosis, (2) patient morbidity or mortality risk assessment, (3) disease outbreak prediction and surveillance, and (4) health policy and planning. However, much of the AI-driven intervention research in global health does not describe ethical, regulatory, or practical considerations required for widespread use or deployment at scale. Despite the field remaining nascent, AI-driven health interventions could lead to improved health outcomes in LMICs. Although some challenges of developing and deploying these interventions might not be unique to these settings, the global health community will need to work quickly to establish guidelines for development, testing, and use, and develop a user-driven research agenda to facilitate equitable and ethical use.
Background:
The PRECISE-DAPT and PARIS risk scores (RSs) were recently developed to help clinicians at individualizing the optimal dual antiplatelet therapy duration (DAPT) after percutaneous coronary intervention (PCI). Nevertheless, external validation of these RSs it has not yet been performed in ACS (acute coronary syndrome) patients treated with prasugrel or ticagrelor in a real- world scenario.
Methods:
4424 ACS patients who underwent PCI and survived to hospital discharge, from January 2012 to December 2016 at 12 European centers, were included. PRECISE-DAPT and PARIS bleeding RS, as well as PARIS ischemic RS, were computed, and their performance at predicting major bleeding (MB; BARC type 3 or 5) and ischemic events (MI and stent thrombosis) during follow up was compared.
Results:
After a median follow-up of 14 (interquartile range 12-20.9) months, 83 (1.88%) patients developed MB and 133 (3.0%) suffered an ischemic episode. PRECISE-DAPT performed better than PARIS bleeding RS (c-statistic = 0.653 vs. 0.593; p = .01 for comparison) in predicting MB. The RSs performance for MB prediction remained consistent in STEMI patients (c-statistic = 0.632 vs 0.575) or in those treated with prasugrel (c-statistic = 0.623 vs 0.586). PARIS ischemic RS exhibited superior discrimination in predicting ischemic complications compared to PRECISE-DAPT (c-statistic = 0.604 vs 0.568 p = .05 for comparison).
Conclusion:
Our data provide support to the use of PRECISE-DAPT in MB risk stratification for patients receiving DAPT in form of aspirin and prasugrel or ticagrelor whereas the PARIS ischemic RS has potential to complement the risk prediction with respect to ischemic events.
Background:
The risk of recurrent ischemia and bleeding after percutaneous coronary intervention (PCI) for acute coronary syndrome (ACS) may vary during the first year of follow-up according to clinical presentation, and medical and interventional strategies.
Methods:
BleeMACS and RENAMI are 2 multicenter registries enrolling patients with ACS treated with PCI and clopidogrel, prasugrel, or ticagrelor. The average daily ischemic and bleeding risks (ADIR and ADBR) in the first year after PCI were the primary end points. The difference between ADBR and ADIR was calculated to estimate the potential excess of bleeding/ischemic events in a given period or specific subgroup.
Results:
A total of 19,826 patients were included. Overall, in the first year after PCI, the ADBR was 0.008085%, whereas ADIR was 0.008017% (P = .886). In the first 2 weeks ADIR was higher than ADBR (P = .013), especially in patients with ST-segment elevation myocardial infarction or incomplete revascularization. ADIR continued to be, albeit non-significantly, greater than ADBR up to the third month, whereas ADBR became higher, although not significantly, afterward. Patients with incomplete revascularization had an excess in ischemic risk (P = .003), whereas non-ST-segment elevation ACS patients and those on ticagrelor had an excess of bleeding (P = .012 and P = .022, respectively).
Conclusions:
In unselected ACS patients, ADIR and ADBR occurred at similar rates within 1 year after PCI. ADIR was greater than ADBR in the first 2 weeks, especially in ST-segment elevation myocardial infarction patients and those with incomplete revascularization. In the first year, ADIR was higher than ADBR in patients with incomplete revascularization, whereas ADBR was higher in non-ST-segment elevation ACS patients and in those discharged on ticagrelor.
Aims:
Aim of the present study was to establish the safety and efficacy profile of prasugrel and ticagrelor in real-life acute coronary syndrome (ACS) patients with renal dysfunction.
Methods and results:
All consecutive patients from RENAMI and BLEEMACS registries were stratified according to estimated glomerular filtration rate (eGFR) lower or greater than 60mL/min/1.73m2. Death and myocardial infarction (MI) were the primary efficacy endpoints. Major bleedings (MB), defined as Bleeding Academic Research Consortium bleeding types 3 to 5, constituted the safety endpoint.19255 patients were enrolled. Mean age was 63 ± 12; 14892 (77.3%) were males. 2490 (12.9%) patients had chronic kidney disease (CKD), defined as eGFR<60mL/min/1.73m2. Mean follow-up was 13±5 months. Mortality was significantly higher in CKD patients (9.4% vs 2.6%, p < 0.0001), as well as the incidence of reinfarction (5.8% vs 2.9%, p < 0.0001) and MB (5.7% vs 3%, p < 0.0001). At Cox multivariable analysis, potent P2Y12 inhibitors significantly reduced the mortality rate (HR 0.82, 95% CI 0.54-0.96, p = 0.006) and the risk of reinfarction (HR 0.53, 95% CI 0.30-0.95, p = 0.033) in CKD patients as compared to clopidogrel. The reduction of risk of re-infarction was confirmed in patients with preserved renal function. Potent P2Y12 inhibitors did not increase the risk of MB in CKD patients (HR 1.00, 95% CI 0.59-1.68, p = 0.985).
Conclusion:
In ACS patients with CKD, prasugrel and ticagrelor are associated with lower risk of death and recurrent MI without increasing the risk of MB.
Aims:
A large trial established the favorable clinical profile of a new polymer-free biolimus-A9-eluting stent (PF-BES) with a 1-month dual antiplatelet therapy (DAPT) regimen in patients at high bleeding risk (HBR). We evaluated the real-world patterns of indications, DAPT strategies and outcomes for the PF-BES following this evidence.
Methods and results:
CHANCE is a multicenter registry including all patients who underwent percutaneous coronary intervention (PCI) with at least one PF-BES. Reasons for PF-BES PCI and planned antithrombotic regimens were collected. Primary outcomes were the 390-day Kaplan Meier estimates of a patient-oriented and a device-oriented composite endpoints (POCE: death, myocardial infarction [MI] or target vessel revascularization [TVR]; DOCE: cardiac death, target vessel-MI or ischemia-driven target lesion revascularization [ID-TLR]). Between January 2016 and July 2018, 858 patients (age: 74 ±10 years, 64.6% males, 58.7% acute coronary syndrome presentation) underwent PF-BES PCI. Main reasons for PF-BES physician's choice reflected a perceived HBR in 77.7% of patients. One-month DAPT was planned in 40.3% of patients. At 390-day follow-up (median 340 days, interquartile range: 187-390 days) the incident estimate of POCE was 13.1% (any MI 3.7%, any TVR 3.4%) and of DOCE was 7.1% (TV-MI 3.6%, ID-TLR 1.4%); while 390-day estimate of any bleeding event was 11.1% (BARC 3-5 bleeding 3.0%).
Conclusions:
In a large all-comers registry, PF-BES was mostly used in HBR patients, frequently followed by very-short DAPT regimen. The reported outcomes suggest a favorable safety and efficacy profile for the PF-BES in a real-world clinical setting. (ClinicalTrials.gov identifier: NCT03622203).
Currently available risk prediction methods are limited in their ability to deal with complex, heterogeneous, and longitudinal data such as that available in primary care records, or in their ability to deal with multiple competing risks. This paper develops a novel deep learning approach that is able to successfully address current limitations of standard statistical approaches such as landmarking and joint modeling. Our approach, which we call Dynamic-DeepHit, flexibly incorporates the available longitudinal data comprising various repeated measurements (rather than only the last available measurements) in order to issue dynamically updated survival predictions for one or multiple competing risk(s). Dynamic-DeepHit learns the time-to-event distributions without the need to make any assumptions about the underlying stochastic models for the longitudinal and the time-to-event processes. Thus, unlike existing works in statistics, our method is able to learn data-driven associations between the longitudinal data and the various associated risks without underlying model specifications. We demonstrate the power of our approach by applying it to a real-world longitudinal dataset from the UK Cystic Fibrosis Registry which includes a heterogeneous cohort of 5,883 adult patients with annual follow-ups between 2009-2015. The results show that Dynamic-DeepHit provides a drastic improvement in discriminating individual risks of different forms of failures due to cystic fibrosis. Furthermore, our analysis utilizes postprocessing statistics that provide clinical insight by measuring the influence of each covariate on risk predictions and the temporal importance of longitudinal measurements, thereby enabling us to identify covariates that are influential for different competing risks.
Introduction:
The benefits of short versus long-term dual antiplatelet therapy (DAPT) based on the third generation P2Y12 antagonists prasugrel or ticagrelor, in patients with acute coronary syndromes treated with percutaneous coronary intervention remain to be clearly defined due to current evidences limited to patients treated with clopidogrel.
Methods:
All acute coronary syndrome patients from the REgistry of New Antiplatelets in patients with Myocardial Infarction (RENAMI) undergoing percutaneous coronary intervention and treated with aspirin, prasugrel or ticagrelor were stratified according to DAPT duration, that is, shorter than 12 months (D1 group), 12 months (D2 group) and longer than 12 months (D3 group). The three groups were compared before and after propensity score matching. Net adverse clinical events (NACEs), defined as a combination of major adverse cardiac events (MACEs) and major bleedings (including therefore all cause death, myocardial infarction and Bleeding Academic Research Consortium (BARC) 3-5 bleeding), were the primary end points, MACEs (a composite of all cause death and myocardial infarction) the secondary one. Single components of NACEs were co-secondary end points, along with BARC 2-5 bleeding, cardiovascular death and stent thrombosis.
Results:
A total of 4424 patients from the RENAMI registry with available data on DAPT duration were included in the model. After propensity score matching, 628 patients from each group were selected. After 20 months of follow up, DAPT for 12 months and DAPT for longer than 12 months significantly reduced the risk of NACE (D1 11.6% vs. D2 6.7% vs. D3 7.2%, p = 0.003) and MACE (10% vs. 6.2% vs. 2.4%, p < 0.001) compared with DAPT for less than 12 months. These differences were driven by a reduced risk of all cause death (7.8% vs. 1.3% vs. 1.6%, p < 0.001), cardiovascular death (5.1% vs. 1.0% vs. 1.2%, p < 0.0001) and recurrent myocardial infarction (8.3% vs. 5.2% vs. 3.5%, p = 0.002). NACEs were lower with longer DAPT despite a higher risk of BARC 2-5 bleedings (4.6% vs. 5.7% vs. 6.2%, p = 0.04) and a trend towards a higher risk of BARC 3-5 bleedings (2.4% vs. 3.3% vs. 3.9%, p = 0.06). These results were not consistent for female patients and those older than 75 years old, due to an increased risk of bleedings which exceeded the reduction in myocardial infarction.
Conclusion:
In unselected real world acute coronary syndrome patients treated with percutaneous coronary intervention, DAPT with prasugrel or ticagrelor prolonged beyond 12 months markedly reduces fatal and non-fatal ischaemic events, offsetting the increased risk deriving from the higher bleeding risk. On the contrary, patients >75 years old and female ones showed a less favourable risk-benefit ratio for longer DAPT due to excess of bleedings.
Objectives:
To evaluate "real life" incidence and independent predictors of major bleeding defined in ACS patients treated with PCI and current standard antithrombotic therapy with prasugrel or ticagrelor.
Methods and results:
The RENAMI project is a multicenter retrospective observational registry enrolling 4424 patients with ACS treated with PCI and prasugrel or ticagrelor plus aspirin. Primary endpoint was MACE (major adverse cardiovascular events). Secondary endpoints included each component of MACE, cardiovascular death (CV death), recurrence of ACS (reACS) and stroke. Eighty three (1.8%) patients developed out of hospital major bleedings after 14.1 ± 6.2 months. These patients had higher rates of MACE (14.5% vs 4.4%; p = 0.001) and of all-cause death (11% vs 2.1%; p < 0.001). Independent predictors of major bleeding were age >75 years (OR 2.00; 95% CI 1.18-3.41; p = 0.010) and female sex (OR 1.66; 95% CI 1.02-2.70; p = 0.041). BARC 3-5 bleeding was independently associated with all-cause mortality (OR 3.46; 95% CI 1.64-7.31; p 0.001).
Conclusion:
In ACS patients treated with PCI and ticagrelor or prasugrel, BARC 3-5 bleedings despite being uncommon negatively impacted on prognosis. Old and female patients are at increased risk, offering clinical indications for tailoring dual antiplatelet therapy.
Introduction and objectives:
The PARIS score allows combined stratification of ischemic and hemorrhagic risk in patients with ischemic heart disease treated with coronary stenting and dual antiplatelet therapy (DAPT). Its usefulness in patients with acute coronary syndrome (ACS) treated with ticagrelor or prasugrel is unknown. We investigated this issue in an international registry.
Methods:
Retrospective multicenter study with voluntary participation of 11 centers in 6 European countries. We studied 4310 patients with ACS discharged with DAPT with ticagrelor or prasugrel. Ischemic events were defined as stent thrombosis or spontaneous myocardial infarction, and hemorrhagic events as BARC (Bleeding Academic Research Consortium) type 3 or 5 bleeding. Discrimination and calibration were calculated for both PARIS scores (PARISischemic and PARIShemorrhagic). The ischemic-hemorrhagic net benefit was obtained by the difference between the predicted probabilities of ischemic and bleeding events.
Results:
During a period of 17.2 ± 8.3 months, there were 80 ischemic events (1.9% per year) and 66 bleeding events (1.6% per year). PARISischemic and PARIShemorrhagic scores were associated with a risk of ischemic events (sHR, 1.27; 95%CI, 1.16-1.39) and bleeding events (sHR, 1.14; 95%CI, 1.01-1.30), respectively. The discrimination for ischemic events was modest (C index = 0.64) and was suboptimal for hemorrhagic events (C index = 0.56), whereas calibration was acceptable for both. The ischemic-hemorrhagic net benefit was negative (more hemorrhagic events) in patients at high hemorrhagic risk, and was positive (more ischemic events) in patients at high ischemic risk.
Conclusions:
In patients with ACS treated with DAPT with ticagrelor or prasugrel, the PARIS model helps to properly evaluate the ischemic-hemorrhagic risk.
ABSTRACT
Introduction: The differential impact on ischemic and bleeding events of the type of drug-eluting stent (DES: durable polymer stents vs. biodegradable polymer stents vs. bioresorbable scaffolds [BRS]) and of the dual antiplatelet therapy (DAPT) duration remains to be defined.
Methods and Results: Randomized controlled trials comparing different types of DES and/or DAPT durations were selected. The primary end-point was MACE (a composite of death, myocardial infarction [MI], and target vessel revascularization). Definite stent thrombosis (ST) and single components of MACE were secondary end-points. The arms of interest were: BRS with 12 months of DAPT (12mDAPT), biodegradable polymer stent with 12mDAPT, durable polymer stent (everolimus-eluting [EES], zotarolimus-eluting [ZES]) with 12mDAPT, EES/ZES with <12 months of DAPT, and EES/ZES with >12 months of DAPT (DAPT>12m). Sixty-four studies with 150 arms and 102,735 patients were included. After a median follow-up of 20 months, MACE rates were similar in the different arms of interest. EES/ZES with DAPT>12m reported a lower incidence of MI than the other groups, while BRS showed a higher rate of ST when compared to EES/ZES, irrespective of DAPT length. A higher risk of major bleedings was observed for DAPT>12m as compared to shorter DAPT.
Conclusion: Durable and biodegradable polymer stents along with BRS report a similar rate of MACE irrespective of DAPT length. Fewer MI are observed with EES/ZES with DAPT>12m, while a higher rate of ST is reported for BRS when compared to EES/ZES, independently from DAPT length. Stent type may partially affect the outcome together with DAPT length.
Background:
Dual antiplatelet therapy (DAPT) with aspirin plus a P2Y12inhibitor prevents ischaemic events after coronary stenting, but increases bleeding. Guidelines support weighting bleeding risk before the selection of treatment duration, but no standardised tool exists for this purpose.
Methods:
A total of 14 963 patients treated with DAPT after coronary stenting-largely consisting of aspirin and clopidogrel and without indication to oral anticoagulation-were pooled at a single-patient level from eight multicentre randomised clinical trials with independent adjudication of events. Using Cox proportional hazards regression, we identified predictors of out-of-hospital Thrombosis in Myocardial Infarction (TIMI) major or minor bleeding stratified by trial, and developed a numerical bleeding risk score. The predictive performance of the novel score was assessed in the derivation cohort and validated in patients treated with percutaneous coronary intervention from the PLATelet inhibition and patient Outcomes (PLATO) trial (n=8595) and BernPCI registry (n=6172). The novel score was assessed within patients randomised to different DAPT durations (n=10 081) to identify the effect on bleeding and ischaemia of a long (12-24 months) or short (3-6 months) treatment in relation to baseline bleeding risk.
Findings:
The PRECISE-DAPT score (age, creatinine clearance, haemoglobin, white-blood-cell count, and previous spontaneous bleeding) showed a c-index for out-of-hospital TIMI major or minor bleeding of 0·73 (95% CI 0·61-0·85) in the derivation cohort, and 0·70 (0·65-0·74) in the PLATO trial validation cohort and 0·66 (0·61-0·71) in the BernPCI registry validation cohort. A longer DAPT duration significantly increased bleeding in patients at high risk (score ≥25), but not in those with lower risk profiles (pinteraction=0·007), and exerted a significant ischaemic benefit only in this latter group.
Interpretation:
The PRECISE-DAPT score is a simple five-item risk score, which provides a standardised tool for the prediction of out-of-hospital bleeding during DAPT. In the context of a comprehensive clinical evaluation process, this tool can support clinical decision making for treatment duration.
Funding:
None.
Aims
Traditional prognostic risk assessment in patients undergoing non-invasive imaging is based upon a limited selection of clinical and imaging findings. Machine learning (ML) can consider a greater number and complexity of variables. Therefore, we investigated the feasibility and accuracy of ML to predict 5-year all-cause mortality (ACM) in patients undergoing coronary computed tomographic angiography (CCTA), and compared the performance to existing clinical or CCTA metrics.
Methods and results
The analysis included 10 030 patients with suspected coronary artery disease and 5-year follow-up from the COronary CT Angiography EvaluatioN For Clinical Outcomes: An InteRnational Multicenter registry. All patients underwent CCTA as their standard of care. Twenty-five clinical and 44 CCTA parameters were evaluated, including segment stenosis score (SSS), segment involvement score (SIS), modified Duke index (DI), number of segments with non-calcified, mixed or calcified plaques, age, sex, gender, standard cardiovascular risk factors, and Framingham risk score (FRS). Machine learning involved automated feature selection by information gain ranking, model building with a boosted ensemble algorithm, and 10-fold stratified cross-validation. Seven hundred and forty-five patients died during 5-year follow-up. Machine learning exhibited a higher area-under-curve compared with the FRS or CCTA severity scores alone (SSS, SIS, DI) for predicting all-cause mortality (ML: 0.79 vs. FRS: 0.61, SSS: 0.64, SIS: 0.64, DI: 0.62; P< 0.001).
Conclusions
Machine learning combining clinical and CCTA data was found to predict 5-year ACM significantly better than existing clinical or CCTA metrics alone.
Dual-antiplatelet therapy with aspirin and clopidogrel after percutaneous coronary intervention reduces the risk for coronary thrombotic events (CTEs) at the expense of increasing risk for major bleeding (MB). Metrics to accurately predict the occurrence of each respective event and inform clinical decision making are lacking.
The aim of this study was to develop and validate separate models to predict risks for out-of-hospital thrombotic and bleeding events after percutaneous coronary intervention with drug-eluting stents.
Using data from 4,190 patients treated with drug-eluting stents and enrolled in the PARIS (Patterns of Non-Adherence to Anti-Platelet Regimen in Stented Patients) registry, separate risk scores were developed to predict CTE (defined as the composite of stent thrombosis or myocardial infarction) and MB (defined as the occurrence of a Bleeding Academic Research Consortium type 3 or 5 bleed). External validation was performed in the ADAPT-DES (Assessment of Dual Antiplatelet Therapy With Drug-Eluting Stents) registry.
Over 2 years, CTEs occurred in 151 patients (3.8%) and MB in 133 (3.3%). Independent predictors of CTEs included acute coronary syndrome, prior revascularization, diabetes mellitus, renal dysfunction, and current smoking. Independent predictors of MB included older age, body mass index, triple therapy at discharge, anemia, current smoking, and renal dysfunction. Each model displayed moderate levels of discrimination and adequate calibration.
Simple risk scores of baseline clinical variables may be useful to predict risks for ischemic and bleeding events after PCI with DES, thereby facilitating clinical decisions surrounding the optimal duration of DAPT. (Patterns of Non-Adherence to Anti-Platelet Regimen in Stented Patients [PARIS]; NCT00998127).
Importance:
Dual antiplatelet therapy after percutaneous coronary intervention (PCI) reduces ischemia but increases bleeding.
Objective:
To develop a clinical decision tool to identify patients expected to derive benefit vs harm from continuing thienopyridine beyond 1 year after PCI.
Design, setting, and participants:
Among 11 648 randomized DAPT Study patients from 11 countries (August 2009-May 2014), a prediction rule was derived stratifying patients into groups to distinguish ischemic and bleeding risk 12 to 30 months after PCI. Validation was internal via bootstrap resampling and external among 8136 patients from 36 countries randomized in the PROTECT trial (June 2007-July 2014).
Exposures:
Twelve months of open-label thienopyridine plus aspirin, then randomized to 18 months of continued thienopyridine plus aspirin vs placebo plus aspirin.
Main outcomes and measures:
Ischemia (myocardial infarction or stent thrombosis) and bleeding (moderate or severe) 12 to 30 months after PCI.
Results:
Among DAPT Study patients (derivation cohort; mean age, 61.3 years; women, 25.1%), ischemia occurred in 348 patients (3.0%) and bleeding in 215 (1.8%). Derivation cohort models predicting ischemia and bleeding had c statistics of 0.70 and 0.68, respectively. The prediction rule assigned 1 point each for myocardial infarction at presentation, prior myocardial infarction or PCI, diabetes, stent diameter less than 3 mm, smoking, and paclitaxel-eluting stent; 2 points each for history of congestive heart failure/low ejection fraction and vein graft intervention; -1 point for age 65 to younger than 75 years; and -2 points for age 75 years or older. Among the high score group (score ≥2, n = 5917), continued thienopyridine vs placebo was associated with reduced ischemic events (2.7% vs 5.7%; risk difference [RD], -3.0% [95% CI, -4.1% to -2.0%], P < .001) compared with the low score group (score <2, n = 5731; 1.7% vs 2.3%; RD, -0.7% [95% CI, -1.4% to 0.09%], P = .07; interaction P < .001). Conversely, continued thienopyridine was associated with smaller increases in bleeding among the high score group (1.8% vs 1.4%; RD, 0.4% [95% CI, -0.3% to 1.0%], P = .26) compared with the low score group (3.0% vs 1.4%; RD, 1.5% [95% CI, 0.8% to 2.3%], P < .001; interaction P = .02). Among PROTECT patients (validation cohort; mean age, 62 years; women, 23.7%), ischemia occurred in 79 patients (1.0%) and bleeding in 37 (0.5%), with a c statistic of 0.64 for ischemia and 0.64 for bleeding. In this cohort, the high-score patients (n = 2848) had increased ischemic events compared with the low-score patients and no significant difference in bleeding.
Conclusion and relevance:
Among patients not sustaining major bleeding or ischemic events 1 year after PCI, a prediction rule assessing late ischemic and bleeding risks to inform dual antiplatelet therapy duration showed modest accuracy in derivation and validation cohorts. This rule requires further prospective evaluation to assess potential effects on patient care, as well as validation in other cohorts.
Trial registration:
clinicaltrials.gov Identifier: NCT00977938.
Background The incidence, predictors, and prognostic impact of post-discharge bleeding (PDB) after percutaneous coronary intervention (PCI) with drug-eluting stent (DES) implantation are unclear.
Objectives This study sought to characterize the determinants and consequences of PDB after PCI.
Methods The prospective ADAPT-DES (Assessment of Dual Antiplatelet Therapy With Drug-Eluting Stents) study was used to determine the incidence and predictors of clinically relevant bleeding events occurring within 2 years after hospital discharge. The effect of PDB on subsequent 2-year all-cause mortality was estimated by time-adjusted Cox proportional hazards regression.
Results Among 8,582 “all-comers” who underwent successful PCI with DES in the ADAPT-DES study, PDB occurred in 535 of 8,577 hospital survivors (6.2%) at a median time of 300 days (interquartile range: 130 to 509 days) post-discharge. Gastrointestinal bleeding (61.7%) was the most frequent source of PDB. Predictors of PDB included older age, lower baseline hemoglobin, lower platelet reactivity on clopidogrel, and use of chronic oral anticoagulation therapy. PDB was associated with higher crude rates of all-cause mortality (13.0% vs. 3.2%; p < 0.0001). Following multivariable adjustment, PDB was strongly associated with 2-year mortality (hazard ratio [HR]: 5.03; p < 0.0001), with an effect size greater than that of post-discharge myocardial infarction (PDMI) (HR: 1.92; p = 0.009).
Conclusions After successful PCI with DES in an unrestricted patient population, PDB is not uncommon and has a strong relationship with subsequent all-cause mortality, greater that that associated with PDMI. Efforts to reduce PDB may further improve prognosis after successful DES implantation. (Assessment of Dual Antiplatelet Therapy With Drug-Eluting Stents [ADAPT-DES]; NCT00638794)
Background: Serum creatinine concentration is widely used as an index of renal function, but this concentration is affected by factors other than glomerular filtration rate (GFR). Objective: To develop an equation to predict GFR from serum creatinine concentration and other factors. Design: Cross-sectional study of GFR, creatinine clearance, serum creatinine concentration, and demographic and clinical characteristics in patients with chronic renal disease. Patients: 1628 patients enrolled in the baseline period of the Modification of Diet in Renal Disease (MDRD) Study, of whom 1070 were randomly selected as the training sample ; the remaining 558 patients constituted the validation sample. Methods: The prediction equation was developed by stepwise regression applied to the training sample. The equation was then tested and compared with other prediction equations in the validation sample. Results: To simplify prediction of GFR, the equation included only demographic and serum variables. Independent factors associated with a lower GFR included a higher serum creatinine concentration, older age, female sex, nonblack ethnicity, higher serum urea nitrogen levels, and lower serum albumin levels (P < 0.001 for all factors). The multiple regression model explained 90.3% of the variance in the logarithm of GFR in the validation sample. Measured creatinine clearance overestimated GFR by 19%, and creatinine clearance predicted by the Cockcroft-Gault formula overestimated GFR by 16%. After adjustment for this overestimation, the percentage of variance of the logarithm of GFR predicted by measured creatinine clearance or the Cockcroft-Gault formula was 86.6% and 84.2%, respectively. Conclusion: The equation developed from the MDRD Study provided a more accurate estimate of GFR in our study group than measured creatinine clearance or other commonly used equations.
Background:
The optimal duration of dual antiplatelet therapy (DAPT) following second-generation drug-eluting stent (DES) implantation is still debated.
Objectives:
The aim of this study was to test the noninferiority of 6 versus 12 months of DAPT in patients undergoing percutaneous coronary intervention with second-generation DES.
Methods:
The SECURITY (Second Generation Drug-Eluting Stent Implantation Followed by Six- Versus Twelve-Month Dual Antiplatelet Therapy) trial was a 1:1 randomized, multicenter, international, investigator-driven, noninferiority study conducted from July 2009 to June 2014. Patients with a stable or unstable angina diagnosis or documented silent ischemia undergoing revascularization with at least 1 second-generation DES were eligible. The primary endpoint was a composite of cardiac death, myocardial infarction (MI), stroke, definite or probable stent thrombosis, or Bleeding Academic Research Consortium (BARC) type 3 or 5 bleeding at 12 months. The main secondary endpoint was a composite of cardiac death, MI, stroke, definite or probable stent thrombosis, or BARC type 2, 3, or 5 bleeding at 12 and 24 months.
Results:
Overall, 1,399 patients were enrolled in the study and randomized to receive 6 months (n = 682) versus 12 months (n = 717) DAPT. The primary composite endpoint occurred, respectively, in 4.5% versus 3.7% (risk difference 0.8%; 95% confidence interval [CI]: -2.4% to 1.7%; p = 0.469) at 12 months. The upper 95% CI limit was lower than the pre-set margin of 2%, confirming the noninferiority hypothesis (p < 0.05). Moreover, no differences were observed in the occurrence of the secondary endpoint at 12 months (5.3% vs. 4.0%, difference: 1.2%; 95% CI: -1.0 to 3.4; p = 0.273) and between 12 and 24 months (1.5% vs. 2.2%, difference: -0.7%; 95% CI: -2.1 to 0.6; p = 0.289). Finally, no differences were observed in definite or probable stent thrombosis at 12 months (0.3% vs. 0.4%; difference: -0.1%; 95% CI: -0.7 to 0.4; p = 0.694) and between 12 and 24 months of follow-up (0.1% vs. 0%; difference: 0.1%; 95% CI: -0.1 to 0.4; p = 0.305).
Conclusions:
In a low-risk population, the noninferiority hypothesis of 6 vs. 12 months DAPT following second-generation DES implantation appears accepted for the incidence of cardiac death, MI, stroke, definite/probable stent thrombosis, and BARC type 3 or 5 bleeding at 12 months. (Second Generation Drug-Eluting Stent Implantation Followed by Six- Versus Twelve-Month Dual Antiplatelet Therapy; NCT00944333).
Methods: The prediction equation was developed by stepwise regression applied to the training sample. The equation was then tested and compared with other pre- diction equations in the validation sample. Results: To simplify prediction of GFR, the equation in- cluded only demographic and serum variables. Indepen- dent factors associated with a lower GFR included a higher serum creatinine concentration, older age, female sex, nonblack ethnicity, higher serum urea nitrogen levels, and lower serum albumin levels (P , 0.001 for all factors). The multiple regression model explained 90.3% of the vari- ance in the logarithm of GFR in the validation sample. Measured creatinine clearance overestimated GFR by 19%, and creatinine clearance predicted by the Cockcroft-Gault formula overestimated GFR by 16%. After adjustment for this overestimation, the percentage of variance of the logarithm of GFR predicted by measured creatinine clear- ance or the Cockcroft-Gault formula was 86.6% and 84.2%, respectively. Conclusion: The equation developed from the MDRD Study provided a more accurate estimate of GFR in our study group than measured creatinine clearance or other commonly used equations.
This study sought to develop a practical risk score to predict the risk of stent thrombosis (ST) after percutaneous coronary intervention (PCI) for acute coronary syndromes (ACS).
ST is a rare, yet feared complication after PCI with stent implantation. A risk score for ST after PCI in ACS can be a helpful tool to personalize risk assessment.
This study represents a patient-level pooled analysis of 6,139 patients undergoing PCI with stent implantation for ACS in the HORIZONS-AMI (Harmonizing Outcomes With Revascularization and Stents in Acute Myocardial Infarction) and ACUITY (Acute Catheterization and Urgent Intervention Triage Strategy) trials who were randomized to treatment with bivalirudin versus heparin plus a glycoprotein IIb/IIIa inhibitor. The cohort was randomly divided into a risk score development cohort (n = 4,093) and a validation cohort (n = 2,046). Cox regression methods were used to identify clinical, angiographic, and procedural characteristics associated with Academic Research Consortium-defined definite/probable ST at 1 year. Each covariate in this model was assigned an integer score based on the regression coefficients.
Variables included in the risk score were type of ACS (ST-segment elevation myocardial infarction, non-ST-segment elevation ACS with ST deviation, or non-ST-segment elevation ACS without ST changes), current smoking, insulin-dependent diabetes mellitus, prior PCI, baseline platelet count, absence of early (pre-PCI) anticoagulant therapy, aneurysmal/ulcerated lesion, baseline TIMI (Thrombolysis In Myocardial Infarction) flow grade 0/1, final TIMI flow grade <3, and number of treated vessels. Risk scores 1 to 6 were considered low risk, 7 to 9 intermediate risk, and 10 or greater high risk for ST. Rates of ST at 1 year in low-, intermediate-, and high-risk categories were 1.36%, 3.06%, and 9.18%, respectively, in the development cohort (p for trend <0.001), and 1.65%, 2.77%, and 6.45% in the validation cohort (p for trend = 0.006). The C-statistic for this risk score was over 0.65 in both cohorts.
The individual risk of ST can be predicted using a simple risk score based on clinical, angiographic, and procedural variables. (Harmonizing Outcomes With Revascularization and Stents in Acute Myocardial Infarction [HORIZONS-AMI]; NCT00433966) (Comparison of Angiomax Versus Heparin in Acute Coronary Syndromes [ACUITY]; NCT00093158).
Advances in antithrombotic therapy, along with an early invasive strategy, have reduced the incidence of recurrent ischemic events and death in patients with acute coronary syndromes (ACS; unstable angina, non–ST-segment–elevation myocardial infarction [MI], and ST-segment–elevation MI).1,–,4 However, the combination of multiple pharmacotherapies, including aspirin, platelet P2Y12 inhibitors, heparin plus glycoprotein IIb/IIIa inhibitors, direct thrombin inhibitors, and the increasing use of invasive procedures, has also been associated with an increased risk of bleeding.
Editorial see p 2664
Bleeding complications have been associated with an increased risk of subsequent adverse outcomes, including MI, stroke, stent thrombosis, and death, in patients with ACS and in those undergoing percutaneous coronary intervention (PCI),5,–,10 as well as in the long-term antithrombotic setting.11,12 Thus, balancing the anti-ischemic benefits against the bleeding risk of antithrombotic agents and interventions is of paramount importance in assessing new therapies and in managing patients. Prior randomized trials comparing antithrombotic agents suggest that a reduction in bleeding events is associated with improved survival.13,14
Because prevention of major bleeding may represent an important step in improving outcomes by balancing safety and efficacy in the contemporary treatment of ACS, bleeding events have been systematically identified as a crucial end point for the assessment of the safety of drugs during the course of randomized clinical trials, and are an important aspect of the evaluation of new devices and interventional therapies.15 Unlike ischemic clinical events (eg, cardiac death, MI, stent thrombosis), for which there is now general consensus on end-point definitions,16,17 there is substantial heterogeneity among the many bleeding definitions currently in use. Lack of standardization makes it difficult to optimally organize key clinical trial processes such as adjudication, and even more difficult to interpret relative …
The aim of this study was to develop a practical risk score to predict the risk and implications of major bleeding in acute coronary syndromes (ACS).
Hemorrhagic complications have been strongly linked with subsequent mortality in patients with ACS.
A total of 17,421 patients with ACS (including non-ST-segment elevation myocardial infarction [MI], ST-segment elevation MI, and biomarker negative ACS) were studied in the ACUITY (Acute Catheterization and Urgent Intervention Triage strategY) and the HORIZONS-AMI (Harmonizing Outcomes with RevasculariZatiON and Stents in Acute Myocardial Infarction) trials. An integer risk score for major bleeding within 30 days was developed from a multivariable logistic regression model.
Non-coronary artery bypass graft surgery (CABG)-related major bleeding within 30 days occurred in 744 patients (7.3%) and had 6 independent baseline predictors (female sex, advanced age, elevated serum creatinine and white blood cell count, anemia, non-ST-segment elevation MI, or ST-segment elevation MI) and 1 treatment-related variable (use of heparin + a glycoprotein IIb/IIIa inhibitor rather than bivalirudin alone) (model c-statistic = 0.74). The integer risk score differentiated patients with a 30-day rate of non-CABG-related major bleeding ranging from 1% to over 40%. In a time-updated covariate-adjusted Cox proportional hazards regression model, major bleeding was an independent predictor of a 3.2-fold increase in mortality. The link to mortality risk was strongest for non-CABG-related Thrombolysis In Myocardial Infarction (TIMI)-defined major bleeding followed by non-TIMI major bleeding with or without blood transfusions, whereas isolated large hematomas and CABG-related bleeding were not significantly associated with subsequent mortality.
Patients with ACS have marked variation in their risk of major bleeding. A simple risk score based on 6 baseline measures plus anticoagulation regimen identifies patients at increased risk for non-CABG-related bleeding and subsequent 1-year mortality, for whom appropriate treatment strategies can be implemented.
P2Y12 inhibitors in acute coronary syndrome patients with renal dysfunction: an analysis from the RENAMI and BleeMACS projects
- O De Filippo
- D 'ascenzo
- F Raposeirasroubin
De Filippo O, D'Ascenzo F, RaposeirasRoubin S, et al.
P2Y12 inhibitors in acute coronary syndrome patients with renal
dysfunction: an analysis from the RENAMI and BleeMACS
projects. Eur Heart J Cardiovasc Pharmacother 2020; 6: 31-42.
Usefulness of the PARIS Score to evaluate the ischemic-hemorrhagic net benefit with ticagrelor and prasugrel after an acute coronary syndrome
- Raposeiras-Roubín