Article

Analysis of heart rate variability as a predictor of mortality in cardiovascular patients of intensive care unit

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

Objective: Dynamic changes of heart rate variability (HRV) reflect autonomic dysfunction in cardiac disease. Some studies suggest the role of HRV in predicting intensive care unit (ICU) mortality. The main object of this study was analyzing the HRV to design an algorithm to predict mortality risk. Methods: We evaluated 80 cardiovascular ICU patients (45 males and 45 females), ranging from 45 to 70 years. Common time and frequency domain analysis, non-linear Poincaré plot and recurrence quantification analysis (RQA) were used to study the HRV in two episodes. The episodes include 8-4. h before death, and 4. h before death to death. Independent sample t-test was used as statistical analysis. Results: Statistical analysis indicates that frequency domain and Poincaré parameters such as LF/HF and SD2/SD1 show changes in transition to death episode (p <0.05). Moreover, L mean , v max and RT measures showed meaningful changes (p <0.01) in closer segments to the death. Conclusions: Analysis of physiological variables shows that there are significant differences in RQA measures in episodes close to death. These changes can be interpreted as more stability and determinism behavior of HRV in episodes close to death. RQA parameters can be used together with HRV parameters for description and prediction of mortality risk in ICU patients. © 2015 Nałęcz Institute of Biocybernetics and Biomedical Engineering.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... The intensive care unit (ICU) is a special place where medical personnel and equipment are employed for the treatment and management of critically ill patients. An acceptable target in this section is to save the lives of survived patients because all patients admitted to the ICU would not return to normal life and perhaps the life, and some patients will die due to the severity of the disease [2]. The intensive care unit should not be considered as a place for the death of patients. ...
... Because heart rate and blood pressure monitoring in the cerebrovascular disease of the intensive care unit is very important, so determining the statistical characteristics of these signals, including mean and variance, can be helpful as an adjunct in analyzing the behavior of the signals in these patients. The features of this area include the standard deviation of NN intervals (SDNN), the standard deviation of the average of NN intervals in all, the root mean square of successive difference (RMSSD) [2]. The frequency components of the HRV are different. ...
... Due to the fact that each patient was being compared to him (her) self, the paired sample t-test was used to determine the significance between the features of two consecutive windows. Equation (2) shows how to do the windowing method and calculate the features of each window to determine the P-value. (2) Where N, the number of samples in each window, M, the number of windows, is the first angle feature generated of M th window, and are the mean and standard deviation of angle feature in M window. ...
Preprint
Full-text available
Background: This article aimed to explore the mortality prediction of cerebrovascular patients in the intensive care unit (ICU) by examining the important signals associated with these patients during different periods of admission in the intensive care unit, which is considered as one of the new topics in the medical field. Several approaches have been proposed for prediction in this area that each of these methods has been able to predict the mortality somewhat, but many of these techniques require the recording of a large amount of data from the patients, where the recording of all data is not possible in most cases; while this article focuses only on the heart rate variability (HRV) and systolic and diastolic blood pressure. Methods: In this paper, using the information obtained from the electrocardiogram (ECG) signal and blood pressure with the help of vital signal processing methods, how to change these signals during the patient's hospitalization will be initially checked. Then, the mortality prediction in patients with cerebral ischemia is evaluated using the features extracted from the return map generated by the signal of heart rate variability and blood pressure. To implement this paper, 80 recorded data from cerebral ischemic patients admitted to the intensive care unit, including ECG signal recording, systolic and diastolic blood pressure, and other physiological parameters are collected. Time of admission and time of death are labeled in all data. Results: The results indicate that the use of the new approach presented in this article can be compared with other methods or leads to better results. The accuracy, specificity, and sensitivity based on the novel features were, respectively, 97.7, 98.9, and 95.4 for cerebral ischemia disease with a prediction horizon of 0.5-1 hours before death. Conclusion: The perspective of the prediction horizons and the patients' length of stay with a new approach was taken into account in this article. The higher the prediction horizon, the nurses or associates of patients have more time to carry out therapeutic measures. To determine the patient's future status and analysis of the ECG signal and blood pressure, at least 7.8 hours of hospitalization is required, which has had a significant reduction compared with other methods.
... Some of the methods which were previously used to detect the death of cardiac patients include physician checkup, examining genetic factors, medical tests, recording vital signals such as ECG and their interpretation, stress test, and magnetic resonance imaging (MRI) from the heart (Moridani et al. 2015;Moridani and Farhadi 2017). ...
... Currently, some studies have focused on on this serious health problem to develop a technique for predicting and preventing SCD, and increase the survival rate by using invasive and non-invasive methods (Acharya et al. 2015a, b, c). Moridani et al. (2015), Moridani and Farhadi (2017) presented the linear and nonlinear features to classify cardiovascular heart disease. Acharya et al. (2015a, b, c) presented an automatic prediction method using normal signal classification and checking 4-min pre-attack in similar patients. ...
Article
Full-text available
Sudden cardiac death (SCD) generally applied to an unpredicted death from a cardiovascular cause in a subject with or without preexisting heart disease. The main goal of this study was analyzing the Electrocardiogram (ECG) signal to design an algorithm to predict SCD risk. In this paper, ECG signals of 23 subjects (13 males, 8 females and 2 unknown), ranging from 17 to 89 years old necessary for the research were obtained from the Physionet database. For this purpose, we developed a new method to predict SCD, a 10-min prior heart attack using the return map. The aim of this study is a novel method based on Lag return map for in control patients and SCD classes. Return map with six different lags (1–6) was constructed in two-time intervals. After that, the non-linear features that include SD1, SD2, SD1/SD2 for each Lag was measured. The result shows that the rate of changes in SD1 and SD1/SD2 with increasing lags were increased significantly but in SD2 with increasing lags was decreased in two intervals. Statistical analysis indicates that return map parameters show changes in the transition to death episode (p < 0.05). Besides, there were significant changes (p < 0.01) in closer segments to death. In conclusion, it will be possible to predict SCD based on the nonlinear feature that can alarm doctors of an imminent SCD, helping them provide timely treatments that can increase the survival rate of patients and thus reduce the mortality rate.
... Although deaths from heart disease have declined over the past 20 years [8], sudden cardiac death is still the cause of half of the deaths from cardiovascular diseases [9]. Early methods for predicting the mortality of heart patients include physician examination, examining genetic factors, medical tests, recording vital signals such as electrocardiogram (ECG) and their interpretation, stress test, magnetic resonance imaging (MRI) from the heart, and so on [10]. Data mining techniques also play a key role in the field of medicine, because they are used to discover, analyze and extract medical data using sophisticated algorithms [11,12]. ...
... The distance between these beats varies depending on the individual's condition at different times. The signal caused by RR interval over time is called the HRV [10]. Fig. 1 shows the image of the ECG signal with the detection of R peaks and the production of the HRV signal. ...
Article
Full-text available
Intensive care unit (ICU) experienced and skillful people in this field should be employed because the equipment, facilities, and admitted patients have more special conditions than other departments. Our goal provides the best quality according to the condition each patient and prevent many unnecessary costs for preventive treatment. In this paper, the proposed system will first receive the patient's vital signs, which are recorded by the ICU monitoring. After the necessary processing, in case of observing changes in the normal state, risk alarms are transmitted to the nursing station so that nurses become aware of this condition and take all equipment to return the patient to normal condition and prevent his death. The applied graph in this study examines patients at any moment and displays the patient's future condition in a schematic manner after precise analyses. In this algorithm, after calculating the R-R intervals in the electrocardiogram signal, RRIs are thrown into a risk plot (RP) by a projectile. Given the amount of projectile RRI, one of the stairs can host that amount. After a few moments by springs embedded under the stairs, the drain of RRIs is done by the kinetic energy stored in the springs towards the valley of life. If the accumulation of quantities in a stair is too much, the spring will not be able to project those RRIs. By examining this situation, we will introduce an index to determine the risk of death for all patients. The results of this paper show that when a person is in normal condition, there is no density in a certain stair and the ball or the projected RRIs are not limited to a stair. In general, the results of this paper show that the lower amount of RRI dispersion in the RP leads to greater risk of entry into the death range and as this amount decrease, an immediate consideration is required. In conclusion, if the precise prediction of the future condition of ICU patients is available to nurses and doctors, more facilities and equipment could be provided to save their lives. • We focused on nonlinear methods with new aspects to extract mentioned dynamics. • This method can reduce the number of ICU nurses and give the special facilities for high-risk patients. • Our results confirm that it is possible to predict mortality based on the dynamical characteristics of HRV.
... Also, some studies have proposed linear and non-linear features to classify cardiovascular heart disease. [23][24][25][26][27][28] But, the method presented in this paper can not only detect the CAD with high accuracy, but also has the ability to visualize the changes in the heart signal and localization of regional electrical activity in the heart due to the use of 12 standard leads. ...
Article
Full-text available
Electrocardiogram (ECG) signals containing very important information about the cardiac are one of the most common tools for physicians in the diagnosis of various types of cardiac diseases. Low accuracy in positioning, limitation of time accuracy, the similarity of signals between some diseases and normal signals and probability of missing some aspect of data are the defect aspects of this method. Importance of cardiac signals and defects of current methods in diagnosis show the need of substituting a new method to show the activity of cardiac. One of the most dangerous defections is ischemia, which corrects and on time diagnose could avoid the latter effect of it. Each of common methods for diagnosis of this illness has their own advantages and disadvantages. In this paper, we consider describing a non-invasive method for ischemic episode detection based on mapping of ECG signals. With this method, we can present the signals with virtual colors and facilitate the diagnosis of ischemic disease. So, a new method of 12-lead cardiac presentation is described that in fact present the 12-lead signals in two images. The result of this paper will present the indicators of sensitivity, specificity and accuracy in the context of disease diagnosis. This paper proposed a novel ECG imaging algorithm for classifying the normal and ischemic signals and 95.35% specificity, 96.79% sensitivity and 95.76% accuracy were achieved which are very much promising compared to the other methods and doctor’s accuracy.
... The selection of a large number of feature functions makes the classifier confused and leads to its failure to distinguish between the two groups of features extracted from the two categories of the signal [32]. Since some of the features are incapable of proper classification, the feature dimensions are reduced with the principal component analysis (PCA) method. ...
Article
Full-text available
p>Heart rate is one of the most important vital signs. People usually face high tension in routine life, and if we found an effective method to control the heart rate, it would be very desirable. One of the goals of this paper is to examine changes in heart rate before and during meditation. Another goal is that what impact could have meditation on the human heartbeat. To heart rate analysis before and during meditation, available heart rate signals have been used for the Physionet database that contains 10 normal subjects and 8 subjects that meditation practice has been done on them. In this paper, first is paid to extract linear and nonlinear characteristics of heart rate and then is paid to the best combination of features to identify two intervals before and during meditation using MLP and SVM classifiers with the help of sensitivity, specificity and accuracy measurements. The achieved results in this paper showed that choosing the best combination of a feature to make a meaningful difference between two intervals before and during meditation includes two-time features (Mean HR, SDNN), a frequency feature ( ), and three nonlinear characteristics ( ). Also, using the support vector machine had better results than the MLP neural network. The sensitivity, specificity, and accuracy of the mean and standard deviation obtained respectively like 92.73 0.23, 89.05 0.67, 89.97 0.23 by using MLP and respectively like 95.96 0.09, 93.80 0.16, and 94.90 0.14 by using SVM. As a result, using meditation can reduce the stress and anxiety of patients by effects on heart rate, and the treatment process speeds up and have an important role in improving the performance of the system.
... Many applications have been introduced to analyze the heart signal in medicine. HRV analysis not only has received much attention in detection, prediction, classification, or even treatment of cardiovascular diseases but also for the understanding of the psychological disorders and states [1][2][3]. To address this goal, several computer-aided algorithms recruited for offline/online monitoring of HRV. ...
Article
Full-text available
Heart rate variability (HRV) analysis has become a widely used tool for monitoring pathological and psychological states in medical applications. In a typical classification problem, information fusion is a process whereby the effective combination of the data can achieve a more accurate system. The purpose of this article was to provide an accurate algorithm for classifying HRV signals in various psychological states. Therefore, a novel feature level fusion approach was proposed. First, using the theory of information, two similarity indicators of the signal were extracted, including correntropy and Cauchy-Schwarz divergence. Applying probabilistic neural network (PNN) and k-nearest neighbor (kNN), the performance of each index in the classification of meditators and non-meditators HRV signals was appraised. Then, three fusion rules, including division, product, and weighted sum rules were used to combine the information of both similarity measures. For the first time, we propose an algorithm to define the weights of each feature based on the statistical p-values. The performance of HRV classification using combined features was compared with the non-combined features. Totally, the accuracy of 100% was obtained for discriminating all states. The results showed the strong ability and proficiency of division and weighted sum rules in the improvement of the classifier accuracies.
... To improve the accuracy of the mortality prediction, researchers have proposed models and scoring systems under specific conditions. For example, Moridani et al. [8] proposed a mortality scoring system for people with heart problems. They concluded that the model provides acceptable results when risks are predicted early. ...
Article
Full-text available
The intensive care unit (ICU) admits the most seriously ill patients requiring extensive monitoring. Early ICU mortality prediction is crucial for identifying patients who are at great risk of dying and for providing suitable interventions to save their lives. Accordingly, early prediction of patients at high mortality risk will enable their provision of appropriate and timely medical services. Although various severity scores and machine-learning models have recently been developed for early mortality prediction, such prediction remains challenging. This paper proposes a novel stacking ensemble approach to predict the mortality of ICU patients. Our approach is more accurate and medically intuitive compared to the literature work. Data were prepared and feature selection was processed under the supervision of the ICU domain expert. The data were split into six modalities based on the expert’s decisions. For the prediction process, a separate classifier was selected for each modality based on the performance of the classifiers. We utilized the most popular and diverse classifiers in the literature, including linear discriminant analysis, decision tree (DT), multilayer perceptron, k-nearest neighbor, and logistic regression (LR). Then, a stacking ensemble classifier was constructed and optimized based on the fusion of these five classifier decisions. The framework was evaluated using 10,664 patients from the medical information mart for intensive care (MIMIC III) benchmark dataset. To predict patient mortality, extensive experiments were conducted using the patients’ time series data of different lengths. For each patient, the first 6, 12, and 24 hours of the first stay were tested. The results indicate that our model outperformed the state-of-the-art approaches in terms of accuracy (94.4%), F1 score (93.7%), precision (96.4%), recall (91.1%), and area under the receiver operator characteristic (ROC) curve (93.3%). These results demonstrate the ability and efficiency of our approach to predict ICU mortality.
... Additional features based on the distance between other QRS complex waves, their amplitudes, and statistical characteristics can be used to obtain additional information from the ECG signal. [23] The analysis of HRV is another non-invasive method to assess the function of the ANS used in various clinical conditions, such as stress recognition, heart attack, epileptic seizure classification and prediction, mortality prediction, etc. [24][25][26][27][28][29][30] HRV reflects the balance between the SANS and PANS. [31,32] HRV may be analyzed based on short-term (~5 minutes), ultra-short-term (<5 minutes), or long-term 24-hour recordings. ...
Article
Full-text available
The aim of this article is to summarize current knowledge about the potential clinical utility of electrocardiogram and heart rate variability measures in patients with four common autoimmune diseases )rheumatoid arthritis, systemic lupus erythematosus, Behcet’s disease, and systemic sclerosis(. PubMed, Embase, and Scopus databases were searched for the terms associated with autoimmune diseases, electrocardiogram, heart rate variability, including controlled vocabulary, when appropriate. Articles published in English were considered. Finally, 20 publications were selected, according to the systematic review protocol. Selected papers examined the direct effect of the mentioned diseases on the heart. The parameters of heart rate variability in time-frequency domain analysis were reported to be decreased in patients with autoimmune diseases in comparison with their control groups. QTd and QTc, which is well-known as a risk factor of sudden cardiac death, are increased in the patient group. In some studies, the correlation between the duration of the disease and its activity has been observed. Others have not reported such association. Heart rate turbulence parameters include turbulence onset, was shown to be increased in systemic lupus erythematosus and systemic sclerosis patients, while turbulence slope was decreased in systemic lupus erythematosus patients. These parameters do not show any significant changes in Behcet’s disease. Patients with autoimmune diseases show abnormal heart rate variability and electrocardiogram parameters, which indicates an autonomic cardiac functional impairment. Measurement of electrocardiogram and heart rate variability parameters can be useful clinical tools for the diagnosis and prediction of some disorders in patients with autoimmune diseases.
... In other words, no significant data are lost by integrating features or deleting some. In addition, the high number of features increases the volume of computations, and it is necessary to choose among many features or to combine many features together for achieving more suitable features with fewer dimensions in many cases [33,34]. In the present study, analyzing the main components was implemented as one of the common methods to select and integrate the feature. ...
Article
Full-text available
Abstract Objective The present study aims to simulate an alarm system for online detecting normal electrocardiogram (ECG) signals from abnormal ECG so that an individual's heart condition can be accurately and quickly monitored at any moment, and any possible serious dangers can be prevented. Materials and methods First, the data from Physionet database were used to analyze the ECG signal. The data were collected equally from both males and females, and the data length varied between several seconds to several minutes. The heart rate variability (HRV) signal, which reflects heart fluctuations in different time intervals, was used due to the low spatial accuracy of ECG signal and its time constraint, as well as the similarity of this signal with the normal signal in some diseases. In this study, the proposed algorithm provided a return map as well as extracted nonlinear features of the HRV signal, in addition to the application of the statistical characteristics of the signal. Then, artificial neural networks were used in the field of ECG signal processing such as multilayer perceptron (MLP) and support vector machine (SVM), as well as optimal features, to categorize normal signals from abnormal ones. Results In this paper, the area under the curve (AUC) of the ROC was used to determine the performance level of introduced classifiers. The results of simulation in MATLAB medium showed that AUC for MLP and SVM neural networks was 89.3% and 94.7%, respectively. Also, the results of the proposed method indicated that the more nonlinear features extracted from the ECG signal could classify normal signals from the patient. Conclusion The ECG signal representing the electrical activity of the heart at different time intervals involves some important information. The signal is considered as one of the common tools used by physicians to diagnose various cardiovascular diseases, but unfortunately the proper diagnosis of disease in many cases is accompanied by an error due to limited time accuracy and hiding some important information related to this signal from the physicians' vision leading to the risks of irreparable harm for patients. Based on the results, designing the proposed alarm system can help physicians with higher speed and accuracy in the field of diagnosing normal people from patients and can be used as a complementary system in hospitals.
... Yu and Chen [16] utilized DWT with some statistical as well as frequency features plus a probabilistic neural network as classifier. Many other researches in the field of heartbeat classification can be found in Karimi Moridani et al. [5], Poorahangaryan et al. [12], Yeh et al. [15]. ...
Chapter
Automatic classification of heartbeat is getting a significant value in today’s medical systems. By implementation of these methods in portable diagnosis devices, many mortal diseases can be realized and cured in primary steps. In this paper two separate classifiers are designed and compared for heartbeat classification. The first strategy profits principal component analysis for feature extraction and neural network for classification whereas the second strategy utilizes discrete wavelet transform (DWT) for feature extraction and neural network (NN) as classifier. The arrhythmias which are investigated here include: normal beats (N), right bundle branch block (RBBB), left bundle branch block (LBBB), ventricular premature contraction (VPC) and paced beat (P). In addition, an output for unspecified signals is considered which devotes to anonymous signals which are not in the above list. The results show that both methods could achieve above 98% accuracy on MIT-BIH database.
... This can be effective in improving the performance of the prediction system, reducing false alarms, increasing the prediction horizon and reducing the patient's initial period of stay to apply the prediction system and achieving the ideal parameters for the prediction algorithm. 10 Given the fact that people's heart rates are affected by a variety of factors and experience increases and decreases and various moment fluctuations during the course of the day and night [58], finding the threshold that can help with the less rate of false alarm will be difficult and many of the fluctuations may be confused with the pre-death interval and given a false alarm. It is suggested that, to find a better threshold, a person's behavioral process can be analyzed at least in general situations that may occur throughout the day, for example, quick movements or running to achieve a better pattern for increasing and decreasing the heart rate before death and other cases, because these fluctuations may vary in terms of slope of the increase and decrease. ...
... Selecting a few features prevents highlighting the properties and state of a signal and also fails to distinguish between two different signals. Selecting a large number of feature functions confuses the classifier and leads to its failure to distinguish between the two groups of features extracted from the two categories of the signal [26][27]. ...
... Selecting a few features prevents highlighting the properties and state of a signal and also fails to distinguish between two different signals. Selecting a large number of feature functions confuses the classifier and leads to its failure to distinguish between the two groups of features extracted from the two categories of the signal [26][27]. ...
Conference Paper
Full-text available
Sleep Apnea Syndrome is one of the most common and dangerous causes of sleep disorder that the suspected patients are tested (examined) by recording various types of vital signals during sleep using polysomnography (PSG). Since human body rhythms have a chaotic and non-linear behavior, the nonlinear analysis of body parameters provides the researchers with valuable information about body behavior during the disease and its comparison with the normal state for a more accurate examination of the diseases. The purpose of this is to diagnose apnea events using linear and nonlinear analyses and combining the EMG, ECG and EEG signals in patients with Obstructive Sleep Apnea (OSA). The research data are obtained by the Physionet database including 25 subjects (21 males and 4 females). After performing the pre-processing phase to remove the noise related to EMG, ECG, EEG and artifact signals based on the corresponding algorithms, the healthy and apnea sleep ranges are separated from one another. Linear and nonlinear analyses in MATLAB environment are performed on signals and conditions which are evaluated in healthy sleep and during sleep apnea at different stages of sleep in patients with OSA by multilayer perceptron classifier. The best result of the proposed algorithm obtained by combining the signals and the specificity, sensitivity and accuracy values are 96.87 ± 1.78, 97.14 ± 2.24 and 98.09 ± 2.15 respectively. The results show that the proposed algorithm can help doctors and nurses as a diagnostic tool with more accuracy than similar techniques.
... In this section, we have reviewed the machine learning models based on these risk factors, especially UCI-ML datasets, because our proposed ensemble model presented in this paper is also based on these datasets. But other types of risk factors for heart disease such as heart rate variability (HRV) [20], blood pressure, plasma lipid, Glu, and UA have also been used in the literature to design prediction models. ...
Chapter
Coronary illness has a wide range of conditions that influences the heart. It is one of the most complex disorders to predict because of the number of components in the body that could prompt it. Distinguishing and anticipating it are challenging for specialists and analysts. If we analyze deaths caused by those that were avoidable those from other reasons in India, the third most prevalent cause of unnecessary death is due to heart attack. Across the globe, number of these types of death may increase to more than 23.6 million by 2030. Around 80% of deaths caused by heart attack occur mainly to younger people. This chapter supports medical specialists in detecting and predicting heart disease by attaining precision levels, as well as in prescribing effective medicine according to the disease findings. Given sensor data, deep learning algorithms are applied along with neural network, random forest, and decision tree classifiers to analyze patients' data to predict heart disease. The experiment shows that the prediction of heart disease has promising results with about 90% accuracy.
Conference Paper
In this paper an effective algorithm for classification of four bunches of abnormalities from the normal electrocardiogram is proposed. The considered arrhythmias include; right bundle branch block (RBBB), left bundle branch block (LBBB), premature ventricular contraction (PVC), and Supraventricular tachycardia (SVT) which accompanied with normal signal are investigated. Two levels of feature extraction are totally used for arrhythmia recognition and after each level, the features are sent to neural network (NN) for classification. At first level of feature extraction, ECG signal is compressed using principal component analysis (PCA) which yields feature vectors. Then feature vector is sent to NN as the input. The results showed that this strategy could appropriately identify three species of arrhythmia with high accuracy but had remarkable errors in classifying SVT and RBBB. To overcome this failure, the linear features of heart rate variation plot (HRV) are extracted and used for training a second NN which acts as second classifier for SVT and RBBB. By applying this hybrid technique a precision of one hundred percent is achieved.
Preprint
Full-text available
There is no doubt that the Blockchain has become an important technology that imposes itself in its use. With the increasing demand for this technology it is necessary to develop and update techniques proposed to deal with other technologies, especially in the field of cyber-security, which represents a vital and important field. This paper discussed the integration of Recurrence Qualitative Analysis (RQA) technology with the blockchain as well as exciting technical details of RQA operation in increasing Blockchain security. This paper found significant improvements, remarkable and differentiated compared to previous methods
Article
Introducción: Previamente se desarrolló una nueva metodología de ayuda diagnóstica para los registros Holter fundamentada en los sistemas dinámicos y la teoría de probabilidad, a partir de la información registrada en 21 horas. Objetivo: Evaluar la capacidad diagnóstica de esta metodología durante 19 horas, comparándola con los resultados convencionales del Holter y con los resultados del método matemático aplicado en 21 horas. Materiales y Métodos: fueron evaluados 80 casos de pacientes mayores a 20 años, 10 con registro Holter normal y 70 diagnosticados de forma convencional con diferentes patologías cardíacas. Se establecieron los rangos para las frecuencias cardíacas y de número de latidos por hora en 21 y 19 horas; luego, se calculó la probabilidad de ocurrencia de estos, lo que permitió diferenciar estados de normalidad y enfermedad aguda a partir de tres parámetros. Se comparó el diagnóstico físico-matemático con el diagnóstico convencional, tomado como Gold Standard. Resultados: De los casos normales, dos presentaron probabilidad menor o igual a 0,217 y ocho probabilidades mayores o igual a 0,304; ningún caso de enfermedad aguda presentó valores con probabilidad menor o igual a 0,217, mientras que todos presentaron valores mayores o iguales a 0,304, tanto para los registros Holter evaluados en 21 como en 19 horas. Conclusiones: Se confirmó la utilidad clínica de la metodología ante una reducción del tiempo de evaluación a 19 horas, obteniendo diagnósticos objetivos con base en la auto-organización matemática del fenómeno.
Chapter
Full-text available
Coronary illness may have an awful scope of divergent conditions that could influence your heart. It is one of the furthermost mind‐boggling disorders to foresee a given number of components in your body that can prompt it. Distinguishing and anticipating it represents a lot of challenge for both specialists and analysts. If we extricate the bereavement caused between the avoidable death and the other death in India, the unnecessary death stands in the 3rd position due to Heart‐attack. Across the globe, these types of deaths can increase to more than 23.6 million by 2030. Around 80% of deaths by heart attack mainly affects the younger generation. This research article not only supports doctor specialists in detecting and foreseeing heart disease by attaining precision levels but also advises the effective medicine for the disease. Given sensor data, the deep learning algorithms are applied along with Neural Network, Random Forest and Decision Tree classifier for analyzing the patient's data to predict heart disease. The experiment shows that prediction of heart disease gives promising results of about 90% accuracy.
Preprint
There is no doubt that the Blockchain has become an important technology that imposes itself in its use. With the increasing demand for this technology it is necessary to develop and update techniques proposed to deal with other technologies, especially in the field of cyber-security, which represents a vital and important field. This paper discussed the integration of Recurrence Qualitative Analysis (RQA) technology with the blockchain as well as exciting technical details of RQA operation in increasing Blockchain security. This paper found significant improvements, remarkable and differentiated compared to previous methods
Article
The advent of Artificial Intelligence (AI) has resulted in development of novel applications in a multitude of fields, such as in Medicine, to aid medical professionals in clinical diagnosis. Specifically, the field of Emergency Medicine has been of immense interest to researchers, with vast untapped potential for AI solutions to improve operational efficiencies and quality of healthcare. Aside from primary healthcare facilities, the Emergency Department serves as the first line of contact to patients, who often present with varying and undifferentiated symptoms. Several challenges faced by clinicians and patients alike, such as waiting times and diagnostic dilemmas, present opportunities for application of AI solutions. In this paper, we aim to summarise the applications of AI in the field of Emergency Medicine by reviewing recent developments in Emergency Department operations and in the clinical management of patients.
Article
Full-text available
Different imaging modalities have been developed over the last several years in order to better characterize the atherosclerotic plaque and attempt to predict those in peril of complication. Specific information such as variations in temperature, plaque stiffness and calcification level is currently being researched as well as biological and chemical markers. Since vulnerable plaques cannot be identified by stress testing or angiography, new modalities such as intravascular ultrasound, intracoronary thermography, intravascular palpography, optical coherence tomography, intravascular radiation detection, magnetic resonance imaging, radio nucleotide imaging, and spectroscopy are under investigation. The ultimate goal is to visualize the plaque and its characteristics, stratify its risk for acute events, and are able to apply this modality to the general population of cardiac patients, while exposing the patient to minimal risk and having adequate positive and negative predictive values. In this paper we consider to analyze and compare the Atherosclerotic Plaque detection methods.
Conference Paper
Full-text available
Recurrence is a fundamental property of dynamical systems, which can be exploited to characterise the system's behaviour in phase space. A powerful tool for their visualisation and analysis is the recurrence plot. Methods basing on recurrence plots have been proven to be very successful especially in analysing short, noisy and nonstationary data, as they are typical in Earth sciences. Recurrence Plots (RPs) have found applications in such diverse fields as life sciences, astrophysics, earth sciences, meteorology, biochemistry, and finance, where they are used to provide measures of dynamical properties, complexity or dynamical transitions. Theoretical results show how closely RPs are linked to dynamical invariants like entropies and dimensions. Moreover, they are successful tools for synchronisation analysis and advanced surrogate tests.
Article
Full-text available
A new graphical tool for measuring the time constancy of dynamical systems is presented and illustrated with typical examples.
Article
Full-text available
Context While the adoption of practice guidelines is standardizing many aspects of patient care, ethical dilemmas are occurring because of forgoing life-sustaining therapies in intensive care and are dealt with in diverse ways between different countries and cultures.Objectives To determine the frequency and types of actual end-of-life practices in European intensive care units (ICUs) and to analyze the similarities and differences.Design and Setting A prospective, observational study of European ICUs.Participants Consecutive patients who died or had any limitation of therapy.Intervention Prospectively defined end-of-life practices in 37 ICUs in 17 European countries were studied from January 1, 1999, to June 30, 2000.Main Outcome Measures Comparison and analysis of the frequencies and patterns of end-of-life care by geographic regions and different patients and professionals.Results Of 31 417 patients admitted to ICUs, 4248 patients (13.5%) died or had a limitation of life-sustaining therapy. Of these, 3086 patients (72.6%) had limitations of treatments (10% of admissions). Substantial intercountry variability was found in the limitations and the manner of dying: unsuccessful cardiopulmonary resuscitation in 20% (range, 5%-48%), brain death in 8% (range, 0%-15%), withholding therapy in 38% (range, 16%-70%), withdrawing therapy in 33% (range, 5%-69%), and active shortening of the dying process in 2% (range, 0%-19%). Shortening of the dying process was reported in 7 countries. Doses of opioids and benzodiazepines reported for shortening of the dying process were in the same range as those used for symptom relief in previous studies. Limitation of therapy vs continuation of life-sustaining therapy was associated with patient age, acute and chronic diagnoses, number of days in ICU, region, and religion (P<.001).Conclusion The limiting of life-sustaining treatment in European ICUs is common and variable. Limitations were associated with patient age, diagnoses, ICU stay, and geographic and religious factors. Although shortening of the dying process is rare, clarity between withdrawing therapies and shortening of the dying process and between therapies intended to relieve pain and suffering and those intended to shorten the dying process may be lacking. Figures in this Article While the principle that dying patients should be treated with respect and compassion is broadly accepted among health care professionals, medical practices for end-of-life care differ around the world. In the United States, medicine has moved from a paternalistic model to one that promotes autonomy and self-determination.1- 2 Patient expectations and preferences now help shape end-of-life practices, limiting the use of technologies that may prolong dying rather than facilitate recovery.1- 2 In Europe, patient-physician relationships are still somewhat paternalistic.3- 5 Different cultures and countries deal in diverse ways with the ethical dilemmas arising as a consequence of the wider availability of life-sustaining therapies.3- 4,6 Some have not adopted the Western emphasis on patient autonomy or methods of terminating life support.3- 4,6 In the past, patients died in intensive care units (ICUs) despite ongoing aggressive therapy.7 Theoretical discussions7 and attitudes of critical care professionals concerning these issues have been reported.4,8- 9 In North America, observational studies documenting physician behavior have noted changes in the modes of patient deaths10- 17 and an earlier abandonment of life-sustaining treatments.10 Although European observational studies have demonstrated withholding or withdrawing of life-sustaining treatments in 6% to 13.5% of patients admitted to the ICU and in 35% to 93% of dying patients, the results have come from individual countries.5,17- 21 The overall incidence of end-of-life practices in European ICUs is not known. Furthermore, no studies have been conducted to date comparing different European countries or regions to verify whether reported variations from questionnaire respondents4 are accurate. End-of-life actions in ICUs include withholding or withdrawing life-sustaining therapies,5,7- 21 and in Europe, community studies have described active euthanasia in the Netherlands22 and Belgium.23 At the time this study was conducted, euthanasia was legally practiced in only 1 European country, but withdrawing life-sustaining therapies was common.5,19,21 It is not known whether regional variations in attitudes toward use of euthanasia has any influence on local ICU practice. The objectives of this large multicenter study were to observe and describe actual end-of-life practices in ICUs of several European countries, to determine their overall incidence, to document variations in the pattern of practice, and to analyze similarities and differences in terms of variables that might explain the findings.
Article
Full-text available
To discuss the current role of data mining and Bayesian methods in biomedicine and heath care, in particular critical care. Bayesian networks and other probabilistic graphical models are beginning to emerge as methods for discovering patterns in biomedical data and also as a basis for the representation of the uncertainties underlying clinical decision-making. At the same time, techniques from machine learning are being used to solve biomedical and health-care problems. With the increasing availability of biomedical and health-care data with a wide range of characteristics there is an increasing need to use methods which allow modeling the uncertainties that come with the problem, are capable of dealing with missing data, allow integrating data from various sources, explicitly indicate statistical dependence and independence, and allow integrating biomedical and clinical background knowledge. These requirements have given rise to an influx of new methods into the field of data analysis in health care, in particular from the fields of machine learning and probabilistic graphical models.
Conference Paper
Full-text available
Epilepsy is characterized by the sudden occurrence of seizures disturbing the epileptic patients. Several nonlinear methods have been suggested to predict the onset of epileptic seizures from intracranial or scalp electroencephalogram (EEG) data as "dynamical similarity index analysis" and the "correlation dimension method". Here, special interest is focussed on the focal seizures prediction from scalp EEG data using a nonlinear method called recurrence quantification analysis (RQA)
Article
Full-text available
Recurrence plots have been advocated as a useful diagnostic tool for the assessment of dynamical time series. We extend the usefulness of this tool by quantifying certain features of these plots which may be helpful in determining embeddings and delays.
Article
Full-text available
Heart failure (HF) is a progressive syndrome that marks the end-stage of heart diseases, and it has a high mortality rate and significant cost burden. In particular, non-adherence of medication in HF patients may result in serious consequences such as hospital readmission and death. This study aims to identify predictors of medication adherence in HF patients. In this work, we applied a Support Vector Machine (SVM), a machine-learning method useful for data classification. Data about medication adherence were collected from patients at a university hospital through self-reported questionnaire. The data included 11 variables of 76 patients with HF. Mathematical simulations were conducted in order to develop a SVM model for the identification of variables that would best predict medication adherence. To evaluate the robustness of the estimates made with the SVM models, leave-one-out cross-validation (LOOCV) was conducted on the data set. THE TWO MODELS THAT BEST CLASSIFIED MEDICATION ADHERENCE IN THE HF PATIENTS WERE: one with five predictors (gender, daily frequency of medication, medication knowledge, New York Heart Association [NYHA] functional class, spouse) and the other with seven predictors (age, education, monthly income, ejection fraction, Mini-Mental Status Examination-Korean [MMSE-K], medication knowledge, NYHA functional class). The highest detection accuracy was 77.63%. SVM modeling is a promising classification approach for predicting medication adherence in HF patients. This predictive model helps stratify the patients so that evidence-based decisions can be made and patients managed appropriately. Further, this approach should be further explored in other complex diseases using other common variables.
Thesis
Full-text available
In this work, different aspects and applications of the recurrence plot analysis are presented. First, a comprehensive overview of recurrence plots and their quantification possibilities is given. New measures of complexity are defined by using geometrical structures of recurrence plots. These measures are capable to find chaos-chaos transitions in processes. Furthermore, a bivariate extension to cross recurrence plots is studied. Cross recurrence plots exhibit characteristic structures which can be used for the study of differences between two processes or for the alignment and search for matching sequences of two data series. The selected applications of the introduced techniques to various kind of data demonstrate their ability. Analysis of recurrence plots can be adopted to the specific problem and thus opens a wide field of potential applications. Regarding the quantification of recurrence plots, chaos-chaos transitions can be found in heart rate variability data before the onset of life threatening cardiac arrhythmias. This may be of importance for the therapy of such cardiac arrhythmias. The quantification of recurrence plots allows to study transitions in brain during cognitive experiments on the base of single trials. Traditionally, for the finding of these transitions the averaging of a collection of single trials is needed. Using cross recurrence plots, the existence of an El Niño/Southern Oscillation-like oscillation is traced in northwestern Argentina 34,000 yrs. ago. In further applications to geological data, cross recurrence plots are used for time scale alignment of different borehole data and for dating a geological profile with a reference data set. Additional examples from molecular biology and speech recognition emphasize the suitability of cross recurrence plots.
Article
Full-text available
The objective of this study was to refine the APACHE (Acute Physiology, Age, Chronic Health Evaluation) methodology in order to more accurately predict hospital mortality risk for critically ill hospitalized adults. We prospectively collected data on 17,440 unselected adult medical/surgical intensive care unit (ICU) admissions at 40 US hospitals (14 volunteer tertiary-care institutions and 26 hospitals randomly chosen to represent intensive care services nationwide). We analyzed the relationship between the patient's likelihood of surviving to hospital discharge and the following predictive variables: major medical and surgical disease categories, acute physiologic abnormalities, age, preexisting functional limitations, major comorbidities, and treatment location immediately prior to ICU admission. The APACHE III prognostic system consists of two options: (1) an APACHE III score, which can provide initial risk stratification for severely ill hospitalized patients within independently defined patient groups; and (2) an APACHE III predictive equation, which uses APACHE III score and reference data on major disease categories and treatment location immediately prior to ICU admission to provide risk estimates for hospital mortality for individual ICU patients. A five-point increase in APACHE III score (range, 0 to 299) is independently associated with a statistically significant increase in the relative risk of hospital death (odds ratio, 1.10 to 1.78) within each of 78 major medical and surgical disease categories. The overall predictive accuracy of the first-day APACHE III equation was such that, within 24 h of ICU admission, 95 percent of ICU admissions could be given a risk estimate for hospital death that was within 3 percent of that actually observed (r2 = 0.41; receiver operating characteristic = 0.90). Recording changes in the APACHE III score on each subsequent day of ICU therapy provided daily updates in these risk estimates. When applied across the individual ICUs, the first-day APACHE III equation accounted for the majority of variation in observed death rates (r2 = 0.90, p less than 0.0001).
Article
Full-text available
Power spectrum analysis of heart rate fluctuations provides a quantitative noninvasive means of assessing the functioning of the short-term cardiovascular control systems. We show that sympathetic and parasympathetic nervous activity make frequency-specific contributions to the heart rate power spectrum, and that renin-angiotensin system activity strongly modulates the amplitude of the spectral peak located at 0.04 hertz. Our data therefore provide evidence that the renin-angiotensin system plays a significant role in short-term cardiovascular control in the time scale of seconds to minutes.
Article
Full-text available
We describe refinements to and new experimental applications of the Data Mining Surveillance System (DMSS), which uses a large electronic health-care database for monitoring emerging infections and antimicrobial resistance. For example, information from DMSS can indicate potentially important shifts in infection and antimicrobial resistance patterns in the intensive care units of a single health-care facility.
Article
Full-text available
The development of APACHE (Acute Physiology and Chronic Health Evaluation) began on a Saturday in late June 1978 when I walked into the intensive care unit (ICU) of George Washington University Hospital in Washington, DC. I had come to Washington in 1972 and, with the exception of a year spent working in the former Soviet Union, had completed all of my internal medicine and critical care training in DC. This Saturday morning, however, was unique. It was my first as an attending physician, the last day of fellowship training being the previous day.
Chapter
Recent advances in understanding critical illness and multisystem organ dysfunction (MODS) have resulted from approaches in which individual components of complicated signaling pathways or structures have been identified, by genetic or biochemical means, and their individual properties determined. MODS is thought to originate from a poorly controlled inflammatory response resulting in cellular dysfunction. Altered cellular function results in macroscopic organ system dysfunction [1, 2]. The sequence of events leading to the persistent inflammatory response remains unclear, but much has been learned about the mediators of this response including effector cells (notably neutrophils, monocytes, fixed tissue macrophages, lymphocytes, and vascular endothelial cells) as well as their products: reactive oxygen and nitrogen metabolites, eicosanoids, cytokines, and chemokines acting in an autocrine, paracrine, or endocrine fashion [3–13].
Chapter
The evolution of intensive care has given rise to a unique challenge in medical taxonomy — to describe and characterize the course of diseases that have no biologic precedent. Intensivists debate the optimal definition of such common disorders as sepsis, the acute respiratory distress syndrome (ARDS), and multiple organ failure (MOF),but the roots of this debate he less in the inherent biologic vagaries of the individual processes than in the fact that they are, at root, iatrogenic creations. Acute lung injury only develops in patients whose death has been forestalled by the mechanical ventilator, while the profound physiologic derangements of overwhelming infection are rapidly lethal in the absence of fluid resuscitation, anti-infective therapy, and the spectrum of supportive measures that the contemporary intensive care unit (ICU) provides. The concept of the multiple organ dysfunction syndrome (MODS) reflects an implicit acceptance that the course of critical illness is defined more by what we as physicians have done to sustain the patient, than by the natural history of the extrinsic clinical disorders that rendered the patient critically ill.
Article
Objective. —To develop and validate a new Simplified Acute Physiology Score, the SAPS II, from a large sample of surgical and medical patients, and to provide a method to convert the score to a probability of hospital mortality.Design and Setting. —The SAPS II and the probability of hospital mortality were developed and validated using data from consecutive admissions to 137 adult medical and/or surgical intensive care units in 12 countries.Patients. —The 13 152 patients were randomly divided into developmental (65%) and validation (35%) samples. Patients younger than 18 years, burn patients, coronary care patients, and cardiac surgery patients were excluded.Outcome Measure. —Vital status at hospital discharge.Results. —The SAPS II includes only 17 variables: 12 physiology variables, age, type of admission (scheduled surgical, unscheduled surgical, or medical), and three underlying disease variables (acquired immunodeficiency syndrome, metastatic cancer, and hematologic malignancy). Goodness-of-fit tests indicated that the model performed well in the developmental sample and validated well in an independent sample of patients (P=.883 and P=.104 in the developmental and validation samples, respectively). The area under the receiver operating characteristic curve was 0.88 in the developmental sample and 0.86 in the validation sample.Conclusion. —The SAPS II, based on a large international sample of patients, provides an estimate of the risk of death without having to specify a primary diagnosis. This is a starting point for future evaluation of the efficiency of intensive care units.(JAMA. 1993;270:2957-2963)
Article
Abstract In this paper, we estimate the errors due to observational noise on the recurrence quantification analysis (RQA). Based on this estimation, we present ways to minimize these errors. We give a criterion to choose the threshold ε needed for the optimal computation,of the recurrence plot (RP). One important point is to show the limits of interpretability of the results of the RQA if it is applied to measured time series. We show that even though the RQA is very susceptible to observational noise, it can yield reliable results for an optimal choice of ε if the noise level is not too high. We apply the results to typical models, such as white noise, the logistic map and the Lorenz system, and to experimental laser data. © 2002 Elsevier Science B.V. All rights reserved. PACS: 07.05.Rm; 07.05.Kf Keywords: Recurrence quantification analysis; Recurrence plots; Stochastic processes; Dynamical processes; Observational noise
Article
Recurrence is a fundamental property of dynamical systems, which can be exploited to characterise the system's behaviour in phase space. A powerful tool for their visualisation and analysis called recurrence plot was introduced in the late 1980's. This report is a comprehensive overview covering recurrence based methods and their applications with an emphasis on recent developments. After a brief outline of the theory of recurrences, the basic idea of the recurrence plot with its variations is presented. This includes the quantification of recurrence plots, like the recurrence quantification analysis, which is highly effective to detect, e.g., transitions in the dynamics of systems from time series. A main point is how to link recurrences to dynamical invariants and unstable periodic orbits. This and further evidence suggest that recurrences contain all relevant information about a system's behaviour. As the respective phase spaces of two systems change due to coupling, recurrence plots allow studying and quantifying their interaction. This fact also provides us with a sensitive tool for the study of synchronisation of complex systems. In the last part of the report several applications of recurrence plots in economy, physiology, neuroscience, earth sciences, astrophysics and engineering are shown. The aim of this work is to provide the readers with the know how for the application of recurrence plot based methods in their own field of research. We therefore detail the analysis of data and indicate possible difficulties and pitfalls.
We introduce a simple method to distinguish permanent from transient non-sinus rhythm immediately after resuscitation in cardiac operations. An ice slush of normal saline is sprayed on the surface of the right atrium around the sinus node. If the arrhythmia is transient, the heart usually regains sinus rhythm; if not, the arrhythmia will most likely be permanent. Although we do not know the mechanism, this technique is easy and has worked well in 162 patients.
Article
In this paper, we estimate the errors due to observational noise on the recurrence quantification analysis (RQA). Based on this estimation, we present ways to minimize these errors. We give a criterion to choose the threshold ε needed for the optimal computation of the recurrence plot (RP). One important point is to show the limits of interpretability of the results of the RQA if it is applied to measured time series. We show that even though the RQA is very susceptible to observational noise, it can yield reliable results for an optimal choice of ε if the noise level is not too high. We apply the results to typical models, such as white noise, the logistic map and the Lorenz system, and to experimental laser data.
Article
In this paper we describe the application of data mining methods for predicting the evolution of patients in an intensive care unit. We discuss the importance of such methods for health care and other application domains of engineering. We argue that this problem is an important but challenging one for the current state of the art data mining methods and explain what improvements on current methods would be useful. We present a promising study on a preliminary data set that demonstrates some of the possibilities in this area.
Article
Operational protocols are a valuable means for quality control. However, developing operational protocols is a highly complex and costly task. We present an integrated approach involving both intelligent data analysis and knowledge acquisition from experts that supports the development of operational protocols. The aim is to ensure high quality standards for the protocol through empirical validation during the development, as well as lower development cost through the use of machine learning and statistical techniques. We demonstrate our approach of integrating expert knowledge with data driven techniques based on our effort to develop an operational protocol for the hemodynamic system. 1. (To appear in "Artificial Intelligence in Medicine", thematic issue on Knowledge-Based Information Management in Intensive Care and Anaesthesia) Morik et al: Knowledge Discovery and Knowledge Validation in Intensive Care 2 of 32 2 Key words operational protocols, online-monitoring, time series a...
Article
Heart rate variability (HRV) analysis is affected by ectopic beats. An efficient method was proposed to deal with the ectopic beats. The method was based on trend correlation of the heart timing signal. Predictor of R-R interval (RRI) value at ectopic beat time was constructed by the weight calculation and the slope estimation of preceding normal RRI. The type of ectopic beat was detected and replaced by the predictor of RRI. The performance of the simulated signal after ectopic correction was tested by the standard value using power spectrum density (PSD) estimation, whereas the results of clinical data with ectopic beats were compared with the adjacent ectopic-free data. The result showed the frequency indexes after ectopy corrected had less error than other methods with the test of simulated signal and clinical data. It indicated our method could improve the PSD estimation in HRV analysis. The method had advantages of high accuracy and real time properties to recover the sinus node modulation.
Article
The autonomic nervous system (ANS) plays an important role in the human response to various internal and external stimuli, which can modify homeostasis, and exerts a tight control on essential functions such as circulation, respiration, thermoregulation and hormonal secretion. ANS dysfunction may complicate the perioperative course in the surgical patient undergoing anesthesia, increasing morbidity and mortality, and, therefore, it should be considered as an additional risk factor during pre-operative evaluation. Furthermore, ANS dysfunction may complicate the clinical course of critically ill patients admitted to intensive care units, in the case of trauma, sepsis, neurologic disorders and cardiovascular diseases, and its occurrence adversely affects the outcome. In the care of these patients, the assessment of autonomic function may provide useful information concerning pathophysiology, risk stratification, early prognosis prediction and treatment strategies. Given the role of ANS in the maintenance of systemic homeostasis, anesthesiologists and intensivists should recognize as critical the evaluation of ANS function. Measurement of heart rate variability (HRV) is an easily accessible window into autonomic activity. It is a low-cost, non-invasive and simple to perform method reflecting the balance of the ANS regulation of the heart rate and offers the opportunity to detect the presence of autonomic neuropathy complicating several illnesses. The present review provides anesthesiologists and intensivists with a comprehensive summary of the possible clinical implications of HRV measurements, suggesting that autonomic dysfunction testing could potentially represent a diagnostic and prognostic tool in the care of patients both in the perioperative setting as well as in the critical care arena.
Article
To report incidence and characteristics of decisions to forgo life-sustaining therapies (DFLSTs) in the 282 ICUs who contributed to the SAPS3 database. We reviewed data on DFLSTs in 14,488 patients. Independent predictors of DFLSTs have been identified by stepwise logistic regression. DFLSTs occurred in 1,239 (8.6%) patients [677 (54.6%) withholding and 562 (45.4%) withdrawal decisions]. Hospital mortality was 21% (3,050/14,488); 36.2% (1,105) deaths occurred after DFLSTs. Across the participating ICUs, hospital mortality in patients with DFLSTs ranged from 80.3 to 95.4% and time from admission to decisions ranged from 2 to 4 days. Independent predictors of decisions to forgo LSTs included 13 variables associated with increased incidence of DFLSTs and 7 variables associated with decrease incidence of DFLST. Among hospital and ICU-related variables, a higher number of nurses per bed was associated with increased incidence of DFLST, while availability of an emergency department in the same hospital, presence of a full time ICU-specialist and doctors presence during nights and week-ends were associated with a decreased incidence of DFLST. This large study identifies structural variables that are associated with substantial variations in the incidence and the characteristics of decisions to forgo life-sustaining therapies.
Article
We used 14 easily measured biologic and clinical variables to develop a simple scoring system reflecting the risk of death in ICU patients. The simplified acute physiology score (SAPS) was evaluated in 679 consecutive patients admitted to eight multidisciplinary referral ICUs in France. Surgery accounted for 40% of admissions. Data were collected during the first 24 h after ICU admission. SAPS correctly classified patients in groups of increasing probability of death, irrespective of diagnosis, and compared favorably with the acute physiology score (APS), a more complex scoring system which has also been applied to ICU patients. SAPS was a simpler and less time-consuming method for comparative studies and management evaluation between different ICUs.
Article
The prognostic implications of alterations in heart rate variability have not been studied in a large community-based population. The first 2 hours of ambulatory ECG recordings obtained on original subjects of the Framingham Heart Study attending the 18th biennial examination were reprocessed to assess heart rate variability. Subjects with transient or persistent nonsinus rhythm, premature beats > 10% of total beats, < 1 hour of recording time, processed time < 50% of recorded time, and those taking antiarrhythmic medications were excluded. The associations between heart rate variability measures and all-cause mortality during 4 years of follow-up were assessed. There were 736 eligible subjects with a mean age (+/- SD) of 72 +/- 6 years. The following five frequency domain measures and three time domain measures were obtained: very-low-frequency power (0.01 to 0.04 Hz), low-frequency power (0.04 to 0.15 Hz), high-frequency power (0.15 to 0.40 Hz), total power (0.01 to 0.40 Hz), the ratio of low-frequency to high-frequency power, the standard deviation of total normal RR intervals, the percentage of differences between adjacent normal RR intervals that are > 50 milliseconds, and the square root of the mean of the squared differences between adjacent normal RR intervals. During follow-up, 74 subjects died. In separate proportional hazards regression analyses that adjusted for relevant risk factors, very-low-frequency power (P < .0001), low-frequency power (P < .0001), high-frequency power (P = .0014), total power (P < .0001), and the standard deviation of total normal RR intervals (P = .0019) were significantly associated with all-cause mortality. When all eight heart rate variability measures were assessed in a stepwise analysis that included other risk factors, low-frequency power entered the model first (P < .0001); thereafter, none of the other measures of heart rate variability significantly contributed to the prediction of all-cause mortality. A 1 SD decrement in low-frequency power (natural log transformed) was associated with 1.70 times greater hazard for all-cause mortality (95% confidence interval of 1.37 to 2.09). The estimation of heart rate variability by ambulatory monitoring offers prognostic information beyond that provided by the evaluation of traditional risk factors.
Article
To develop and validate a new Simplified Acute Physiology Score, the SAPS II, from a large sample of surgical and medical patients, and to provide a method to convert the score to a probability of hospital mortality. The SAPS II and the probability of hospital mortality were developed and validated using data from consecutive admissions to 137 adult medical and/or surgical intensive care units in 12 countries. The 13,152 patients were randomly divided into developmental (65%) and validation (35%) samples. Patients younger than 18 years, burn patients, coronary care patients, and cardiac surgery patients were excluded. Vital status at hospital discharge. The SAPS II includes only 17 variables: 12 physiology variables, age, type of admission (scheduled surgical, unscheduled surgical, or medical), and three underlying disease variables (acquired immunodeficiency syndrome, metastatic cancer, and hematologic malignancy). Goodness-of-fit tests indicated that the model performed well in the developmental sample and validated well in an independent sample of patients (P = .883 and P = .104 in the developmental and validation samples, respectively). The area under the receiver operating characteristic curve was 0.88 in the developmental sample and 0.86 in the validation sample. The SAPS II, based on a large international sample of patients, provides an estimate of the risk of death without having to specify a primary diagnosis. This is a starting point for future evaluation of the efficiency of intensive care units.
Article
Sympathetic and parasympathetic activity was evaluated on 39 occasions in 17 patients with the sepsis syndrome, by measurement of the variation in resting heart rate using frequency spectrum analysis. Heart rate was recorded by electrocardiography and respiratory rate by impedance plethysmography. The sepsis syndrome was established on the basis of established clinical and physiological criteria. Subjects were studied, whenever possible, during the period of sepsis and during recovery. Spectral density of the beat-to-beat heart rate was measured within the low frequency band 0.04 to 0.10 Hz (low frequency power, LFP) modulated by sympathetic and parasympathetic activity, and within a 0.12 Hz band width at the respiratory frequency mode (respiratory frequency power, RFP) modulated by parasympathetic activity. Results were expressed as the total variability (total area beneath the power spectrum), as the spectral components normalized to the total power (LFPn, RFPn) or as the ratio of LFP/RFP. During the sepsis syndrome, total heart rate variability and the sympathetically mediated component, LFPn were significantly lower than during the following recovery phase (ANOVA, p < 0.0001, p < 0.01 respectively). Both APACHE II (Acute Physiological and Chronic Health Evaluation) and TISS (Therapeutic Intervention Scoring System) scores showed an inverse correlation with total heart rate variability, logLFP, LFPn and the LFP/RFP ratio (p < 0.002 to 0.0001). Sympathetically mediated heart rate variability was significantly lower during the sepsis syndrome and was inversely proportional to disease severity.
Article
The mutual information I is examined for a model dynamical system and for chaotic data from an experiment on the Belousov-Zhabotinskii reaction. An N logN algorithm for calculating I is presented. As proposed by Shaw, a minimum in I is found to be a good criterion for the choice of time delay in phase-portrait reconstruction from time-series data. This criterion is shown to be far superior to choosing a zero of the autocorrelation function.
Article
We examine the issue of determining an acceptable minimum embedding dimension by looking at the behavior of near neighbors under changes in the embedding dimension from {ital d}{r arrow}{ital d}+1. When the number of nearest neighbors arising through projection is zero in dimension {ital d}{sub {ital E}}, the attractor has been unfolded in this dimension. The precise determination of {ital d}{sub {ital E}} is clouded by noise,'' and we examine the manner in which noise changes the determination of {ital d}{sub {ital E}}. Our criterion also indicates the error one makes by choosing an embedding dimension smaller than {ital d}{sub {ital E}}. This knowledge may be useful in the practical analysis of observed time series.
Article
To study and compare the mode of death in two different institutions' intensive care units (ICUs) for the two time periods, 1988 and 1993. Retrospective chart review. Medical/surgical/trauma ICUs in two tertiary care teaching hospitals. Patients dying in the medical/surgical/trauma ICUs between January 1, 1988 and December 31, 1988; and January 1, 1993 and December 31, 1993. Data collection included demographics, origin of admission, date of ICU admission, date of death, Acute Physiology and Chronic Health Evaluation (APACHE) III diagnostic categories, APACHE II physiologic variables, organ system failures present at the time of admission and 24 hrs before death, and mode of dying. APACHE II scores and mortality risk were calculated. Data analysis included a multiple analysis of variance to assess overall effect, with subsequent analyses of variance to assess the effect of institution and year on each individual dependent variable. All results are reported as mean +/- SEM values. A total of 439 charts were reviewed. Gender, age, and origin of admission were not different between the 2 yrs or the two institutions. Mean APACHE II scores and organ system failures were lower at Hospital A in 1998 vs. Hospital B, as was predicted mortality. These factors increased at Hospital A in 1993 and were similar to those at Hospital B. Withdrawal of support was much more common in 1993 than 1988 at both institutions (43% at Hospital A and 46% at Hospital B in 1988 vs. 66% at A and 80% at B in 1993), increasing to a greater extent in 1993 at Hospital B (p<.05). Length of stay in the ICU was significantly longer at Hospital A than at Hospital B in 1988 (9.4+/-1.4 vs. 4.3+/-0.6 days; p<.05) and in 1993 (8.2+/-2.9 vs. 3.8+/-0.5 days; p < .05). There has been an increase in withdrawal of life support, in recent years, at both the institutions studied. Differences exist between institutions with respect to end-of-life decisions in the ICU. These differences are likely representative of widely prevalent regional differences and are the result of many factors.
Article
Operational protocols are a valuable means for quality control. However, developing operational protocols is a highly complex and costly task. We present an integrated approach involving both intelligent data analysis and knowledge acquisition from experts that support the development of operational protocols. The aim is to ensure high quality standards for the protocol through empirical validation during the development, as well as lower development cost through the use of machine learning and statistical techniques. We demonstrate our approach of integrating expert knowledge with data driven techniques based on our effort to develop an operational protocol for the hemodynamic system.
Article
The objectives of this article are to introduce and explore a novel paradigm based on complex nonlinear systems, and to evaluate its application to critical care research regarding the systemic host response and multiple organ dysfunction syndrome (MODS). Published original work, review articles, scientific abstracts and books, as well as our personal files. Studies were selected for their relevance to the applications of nonlinear complex systems, to critical care medicine, and to the concepts presented. We extracted all applicable data. Following a brief review of MODS, an introduction to complex nonlinear systems is presented, including clear concepts, definitions, and properties. By examining the multiple, nonlinear, interrelated, and variable interactions between the metabolic, neural, endocrine, immune, and inflammatory systems; data regarding interconnected antibody networks; and the redundant, nonlinear, interdependent nature of the inflammatory response, we present the hypothesis that the systemic host response to trauma, shock, or sepsis must be evaluated as a complex nonlinear system. This model provides a new explanation for the failure of trials using various antimediator therapies in the treatment of patients with sepsis and MODS. Understanding the host response as a complex nonlinear system offers innovative means of studying critical care patients, specifically by suggesting a greater focus on systemic properties. We hypothesize that analysis of variability and connectivity of individual variables offer a novel means of evaluating and differentiating the systemic properties of a complex nonlinear system. Current applications of evaluating variability and connectivity are discussed, and insights regarding future research are offered. The paradigm offered by the study of complex nonlinear systems suggests new insights to pursue research to evaluate, monitor, and treat patients with MODS.
Article
In France, there are no guidelines available on withholding and withdrawal of life-sustaining treatments, and information on the frequency of such decisions is scarce. We undertook a prospective 2-month survey in 113, of a total of 220, intensive-care units (ICUs) in France to study the frequency of, and processes leading to, decisions to withhold and withdraw life-sustaining treatments. Life-supporting therapies were withheld or withdrawn in 807 (11.0%) of 7309 patients (withholding in 336 [4.6%] and withdrawal in 471 [6.4%], preceded in 358 by withholding). Of 1175 deaths in ICU, 628 (53%) were preceded by a decision to limit life-supporting therapies. Futility and poor expected quality of life were the most frequently cited reasons. Decisions were strongly correlated with the simplified acute physiological score, but an independent centre effect persisted after adjustment for this score. Decisions were mostly taken by all the ICU medical staff, with (54%) or without (34%) the nursing staff; however, a single physician made decisions in 12% of cases. The patient's family was involved in the decision-making process in 44% of cases. The patient's willingness to limit his or her own care was known in only 8% of the cases; only 0.5% of the patients were involved in decisions. Withholding and withdrawal of life-support therapies are widely practised in French ICUs, despite their prohibition by the French legislation. The lack of an official statement from French scientific bodies may explain several limitations on the various steps of the decision-making process.
Article
Combining the predictions of a set of classifiers has shown to be an effective way to create composite classifiers that are more accurate than any of the component classifiers. There are many methods for combining the predictions given by component classifiers. We introduce a new method that combine a number of component classifiers using a Bayesian network as a classifier system given the component classifiers predictions. Component classifiers are standard machine learning classification algorithms, and the Bayesian network structure is learned using a genetic algorithm that searches for the structure that maximises the classification accuracy given the predictions of the component classifiers. Experimental results have been obtained on a datafile of cases containing information about ICU patients at Canary Islands University Hospital. The accuracy obtained using the presented new approach statistically improve those obtained using standard machine learning methods.
Article
Heart rate variability (HRV) is concerned with the analysis of the intervals between heartbeats. An emerging analysis technique is the Poincaré plot, which takes a sequence of intervals and plots each interval against the following interval. The geometry of this plot has been shown to distinguish between healthy and unhealthy subjects in clinical settings. The Poincaré plot is a valuable HRV analysis technique due to its ability to display nonlinear aspects of the interval sequence. The problem is, how do we quantitatively characterize the plot to capture useful summary descriptors that are independent of existing HRV measures? Researchers have investigated a number of techniques: converting the two-dimensional plot into various one-dimensional views; the fitting of an ellipse to the plot shape; and measuring the correlation coefficient of the plot. We investigate each of these methods in detail and show that they are all measuring linear aspects of the intervals which existing HRV indexes already specify. The fact that these methods appear insensitive to the nonlinear characteristics of the intervals is an important finding because the Poincaré plot is primarily a nonlinear technique. Therefore, further work is needed to determine if better methods of characterizing Poincaré plot geometry can be found.
Article
To determine how frequently life support is withheld or withdrawn from adult critically ill patients, and how physicians and patients families agree on the decision regarding the limitation of life support. Prospective multi-centre cohort study. Six adult medical-surgical Spanish intensive care units (ICUs). Three thousand four hundred ninety-eight consecutive patients admitted to six ICUs were enrolled. Data collected included age, sex, SAPS II score on admission and within 24 h of the decision to limit treatment, length of ICU stay, outcome at ICU discharge, cause and mode of death, time to death after the decision to withhold or withdraw life support, consultation and agreement with patient's family regarding withholding or withdrawal, and the modalities of therapies withdrawn or withheld. Two hundred twenty-six (6.6%) of 3,498 patients had therapy withheld or withdrawn and 221 of them died in the ICU. Age, SAPS II and length of ICU stay were significantly higher in patients dying patients who had therapy withheld or withdrawn than in patients dying despite active treatment. The proposal to withhold or withdraw life support was initiated by physicians in 210 (92.9%) of 226 patients and by the family in the remaining cases. The patient's family was not involved in the decision to withhold or withdraw life support therapy in 64 (28.3%) of 226 cases. Only 21 (9%) patients had expressed their wish to decline life-prolonging therapy prior to ICU admission. The withholding and withdrawing of treatment was frequent in critically ill patients and was initiated primarily by physicians.
Article
We propose a way to automatically detect the best neighborhood size for a local projective noise reduction filter, where a typical problem is the proper identification of the noise level. Here we make use of concepts from the recurrence quantification analysis in order to adaptively tune the filter along the incoming time series. We define an index, to be computed via recurrence plots, whose minimum gives a clear indication of the best size of the neighborhood in the embedding space. Comparison of the local projective noise reduction filter using this optimization scheme with the state of the art is also provided.
Article
We present a case study of machine learning and data mining in intensive care medicine. In the study, we compared different methods of measuring pressure-volume curves in artificially ventilated patients suffering from the adult respiratory distress syndrome (ARDS). Our aim was to show that inductive machine learning can be used to gain insights into differences and similarities among these methods. We defined two tasks: the first one was to recognize the measurement method producing a given pressure-volume curve. This was defined as the task of classifying pressure-volume curves (the classes being the measurement methods). The second was to model the curves themselves, that is, to predict the volume given the pressure, the measurement method and the patient data. Clearly, this can be defined as a regression task. For these two tasks, we applied C5.0 and CUBIST, two inductive machine learning tools, respectively. Apart from medical findings regarding the characteristics of the measurement methods, we found some evidence showing the value of an abstract representation for classifying curves: normalization and high-level descriptors from curve fitting played a crucial role in obtaining reasonably accurate models. Another useful feature of algorithms for inductive machine learning is the possibility of incorporating background knowledge. In our study, the incorporation of patient data helped to improve regression results dramatically, which might open the door for the individual respiratory treatment of patients in the future.
Article
During the past 20 years, ICU risk-prediction models have undergone significant development, validation, and refinement. Among the general ICU severity of illness scoring systems, the Acute Physiology and Chronic Health Evaluation (APACHE), Mortality Prediction Model (MPM), and the Simplified Acute Physiology Score (SAPS) have become the most accepted and used. To risk-adjust patients with longer, more severe illnesses like sepsis and acute respiratory distress syndrome, several models of organ dysfunction or failure have become available, including the Multiple Organ Dysfunction Score (MODS), the Sequential Organ Failure Assessment (SOFA), and the Logistic Organ Dysfunction Score (LODS). Recent innovations in risk adjustment include automatic physiology and diagnostic variable retrieval and the use of artificial intelligence. These innovations have the potential of extending the uses of case-mix and severity-of-illness adjustment in the areas of clinical research, patient care, and administration. The challenges facing intensivists in the next few years are to further develop these models so that they can be used throughout the IUC stay to assess quality of care and to extend them to more specific patient groups such as the elderly and patients with chronic ICU courses.
Article
To examine the frequency and the decision-making processes involved in limiting (withdrawing and withholding) life support therapy in critically ill Chinese patients in the intensive care unit. Prospective survey of patients who had life support limited between April 1997 and March 1999. Medical and surgical intensive care unit of a teaching hospital. All patients admitted to the intensive care unit of the Prince of Wales Hospital who subsequently died and/or had life support limited. Brain-dead patients were excluded from analysis. None. Of 490 patients who died in the intensive care unit, limitation of life support occurred in 288 (58.8%). Relatives or patients requested limitation of life support in 32 cases (11%). The family and/or patient concurred with limitation of life support in 273 occasions (95%). Therapy was withheld in 30.8% and withdrawn in 28.0% of deaths. Therapy limited included inotropes, additional oxygen, and renal replacement therapy. Limitation of therapy in dying Chinese patients occurs frequently in intensive care patients, and both patients and relatives concur with medical decisions to limit therapy in these patients. Withholding therapy rather than withdrawing therapy occurs more frequently than in Western populations.
Article
Decisions made in critical care are often complicated, requiring an in-depth understanding of the relations between complex diseases, available interventions, and patients with a wide range of characteristics. Standard modeling techniques such as decision trees and statistical modeling have difficulty in capturing these interactions as the complexity of the problem increases. Recent models in the literature suggest that simulation modeling techniques such as Markov modeling, Monte Carlo simulation, and discrete-event simulation are useful tools for analyzing complex systems in critical care. These simulation techniques are reviewed briefly, and examples from the literature are presented to demonstrate their usefulness in understanding real problems in critical care. Simulation models provide useful tools for organizing and analyzing the interactions between therapies, tradeoffs, and outcomes.
Article
To review some of the major advances in statistical methodology of the past two decades and their application to investigations in critical care. The introduction of new technologies and the ready availability of advanced computer resources have led to significant developments in statistical methodology. These include the development of computationally intensive methods, improved methods for modeling correlated outcome data, and new methods for handling missing data. Although many of these tools are available in standard statistical software, they remain underutilized in the medical literature. By becoming familiar with advances in statistical methodology, researchers and clinicians can enhance collaboration with their statistical colleagues, toward the goal of better study design and analysis.
Article
To evaluate the implementation and process of withholding and withdrawing life-sustaining treatment in an intensive care unit. Prospective observational study in the medical intensive care unit of a university hospital in Lebanon. Forty-five consecutive adult patients admitted to the ICU for a 1-year period and for whom a decision to withholding and withdrawal of life-sustaining treatment was made. Patients were followed up until their death. Data regarding all aspects of the implementation and the process of withholding and withdrawal of life-sustaining treatment were recorded by a senior staff nurse. Withholding and withdrawing life-sustaining treatment was applied to 9.6% of all admitted patients to ICU. Therapies were withheld in 38% and were withdrawn in 7% of patients who died. Futility of care and poor quality of life were the two most important factors supporting these decisions. The nursing staff was not involved in 26% of the decisions to limit care. Families were not implicated in 21% of the cases. Decisions were not notified in the patients' medical record in 23% of the cases. Sixty-three percent of patients did not have a sedative or an analgesic to treat discomfort during end-of-life care. Life-sustaining treatment were frequently withheld or withdrawn from adult patients in the Lebanese ICU. Cultural differences and the lack of guidelines and official statements could explain the ethical limitations of the decision-making process recorded in this study.