• Toronto, Ontario, Canada
Recent publications
Ingestion of amatoxin-containing mushrooms can lead to fulminant hepatotoxicity and death. Despite recent interest in therapeutic options for amatoxin-exposed patients, there is no single, recommended treatment for amatoxin poisoning. Alpha-amanitin, the principal toxin in amatoxin-containing mushrooms, requires entry into hepatocytes via organic anion transporting polypeptide (OATP) 1B3 before the cascade of cellular events occurs leading to hepatocyte necrosis, liver failure, and in severe cases liver transplantation or death. Three patients managed through a single poison centre with confirmed amatoxin-containing mushrooms ingestions were treated with intravenous cyclosporine, a known potent OATP1B3 inhibitor, along with supportive care. All patients presented with classic delayed symptoms of nausea, vomiting and diarrhea. No patient progressed to fulminant hepatic failure, although two patients developed a transient rise in liver transaminases. All recovered and were discharged from hospital. No patient had an adverse effect from cyclosporine. In addition, we performed an in-vitro study of the role of cyclosporine in cultured HEK293T cells and human hepatoma Huh7 cells. Cyclosporine effectively inhibited OATP1B3-mediated uptake of alpha-amanitin, and improved cell viability of alpha-amanitin exposed cultured Huh7 cells. We conclude that IV cyclosporine, a drug readily available in most hospitals, may be useful to reduce hepatotoxicity from amatoxin poisoning.
Background Chronic obstructive pulmonary disease (COPD) causes 3 million deaths each year, yet 38% of COPD patients continue to smoke. Despite proof of effectiveness and universal guideline recommendations, smoking cessation interventions are underused in practice. We sought to develop an infographic featuring personalized biomedical risk assessment through future lung function decline prediction (with vs without ongoing smoking) to both prompt and enhance clinician delivery of smoking cessation advice and pharmacotherapy, and augment patient motivation to quit. Methods We recruited patients with COPD and pulmonologists from a quaternary care center in Toronto, Canada. Infographic prototype content and design was based on best evidence. After face validation, the prototype was optimized through rapid-cycle design. Each cycle consisted of: (1) infographic testing in a moderated focus group and a clinician interview (recorded/transcribed) (with questionnaire completion); (2) review of transcripts for emergent/critical findings; and (3) infographic modifications to address findings (until no new critical findings emerged). We performed iterative transcript analysis after each cycle and a summative qualitative transcript analysis with quantitative (descriptive) questionnaire analysis. Results Stopping criteria were met after 4 cycles, involving 20 patients (58% male) and 4 pulmonologists (50% male). The following qualitative themes emerged: Tool content (infographic content preferences); Tool Design (infographic design preferences); Advantages of Infographic Messaging (benefits of an infographic over other approaches); Impact of Tool on Determinants of Smoking Cessation Advice Delivery (impact on barriers and enablers to delivery of smoking cessation advice in practice); and Barriers and Enablers to Quitting (impact on barriers and enablers to quitting). Patient Likert scale ratings of infographic content and format/usability were highly positive, with improvements in scores for 20/21 questions through the design process. Providers scored the infographic at 77.8% (“superior”) on the Suitability Assessment of Materials questionnaire. Conclusions We developed a user preference-based personalized biomedical risk assessment infographic to drive smoking cessation in patients with COPD. Our findings suggest that this tool could impact behavioural determinants of provider smoking-cessation advice delivery, while increasing patient quit motivation. Impacts of the tool on provider care, patient motivation to quit, and smoking cessation success should now be evaluated in real-world settings.
Copy number variants (CNVs) are recognized as a crucial genetic cause of neurodevelopmental disorders (NDDs). Chromosomal microarray analysis (CMA), the first-tier diagnostic test for individuals with NDDs, has been utilized to detect CNVs in clinical practice, but most reports are still from populations of European ancestry. To contribute more worldwide clinical genomics data, we investigated the genetic etiology of 410 Han Chinese patients with NDDs (151 with autism and 259 with unexplained intellectual disability (ID) and developmental delay (DD)) using CMA (Affymetrix) after G-banding karyotyping. Among all the NDD patients, 109 (26.6%) carried clinically relevant CNVs or uniparental disomies (UPDs), and 8 (2.0%) had aneuploidies (6 with trisomy 21 syndrome, 1 with 47,XXY, 1 with 47,XYY). In total, we found 129 clinically relevant CNVs and UPDs, including 32 CNVs in 30 ASD patients, and 92 CNVs and 5 UPDs in 79 ID/DD cases. When excluding the eight patients with aneuploidies, the diagnostic yield of pathogenic and likely pathogenic CNVs and UPDs was 20.9% for all NDDs (84/402), 3.3% in ASD (5/151), and 31.5% in ID/DD (79/251). When aneuploidies were included, the diagnostic yield increased to 22.4% for all NDDs (92/410), and 33.6% for ID/DD (87/259). We identified a de novo CNV in 14.9% (60/402) of subjects with NDDs. Interestingly, a higher diagnostic yield was observed in females (31.3%, 40/128) compared to males (16.1%, 44/274) for all NDDs ( P = 4.8 × 10 ⁻⁴ ), suggesting that a female protective mechanism exists for deleterious CNVs and UPDs.
Objectives The objective of this study is to analyse retrospective, observational, longitudinal growth (weight, height and BMI) data in ambulatory boys aged 5–12 years with Duchenne muscular dystrophy (DMD). Background We considered glucocorticoids (GC) use, dystrophin isoforms and amenability to exon 8, 44, 45, 51 and 53 skipping drug subgroups, and the impact of growth on loss of ambulation. We analysed 598 boys, with 2604 observations. This analysis considered patients from the UK NorthStar database (2003–2020) on one of five regimes: “GC naïve”, “deflazacort daily” (DD), “deflazacort intermittent” (DI), “prednisolone daily” (PD) and “prednisolone intermittent” (PI). A random slope model was used to model the weight, height and BMI SD scores (using the UK90). Results The daily regime subgroups had significant yearly height stunting compared to the GC naïve subgroup. Notably, the average height change for the DD subgroup was 0.25 SD (95% CI − 0.30, − 0.21) less than reference values. Those with affected expression of Dp427, Dp140 and Dp71 isoforms were 0.77 (95% CI 0.3, 1.24) and 0.82 (95% CI 1.28, 0.36) SD shorter than those with Dp427 and/or Dp140 expression affected respectively. Increased weight was not associated with earlier loss of ambulation, but taller boys still ambulant between the age of 10 and 11 years were more at risk of losing ambulation. Conclusion These findings may provide further guidance to clinicians when counselling and discussing GCs commencement with patients and their carers and may represent a benchmark set of data to evaluate the effects of new generations of GC.
Current clinical note-taking approaches cannot capture the entirety of information available from patient encounters and detract from patient-clinician interactions. By surveying healthcare providers’ current note-taking practices and attitudes toward new clinical technologies, we developed a patient-centered paradigm for clinical note-taking that makes use of hybrid tablet/keyboard devices and artificial intelligence (AI) technologies. PhenoPad is an intelligent clinical note-taking interface that captures free-form notes and standard phenotypic information via a variety of modalities, including speech and natural language processing techniques, handwriting recognition, and more. The output is unobtrusively presented on mobile devices to clinicians for real-time validation and can be automatically transformed into digital formats that would be compatible with integration into electronic health record systems. Semi-structured interviews and trials in clinical settings rendered positive feedback from both clinicians and patients, demonstrating that AI-enabled clinical note-taking under our design improves ease and breadth of information captured during clinical visits without compromising patient-clinician interactions. We open source a proof-of-concept implementation that can lay the foundation for broader clinical use cases.
Patient engagement in research, a collaborative practice of including patients and families as active and respected partners in the research process, leads to improved quality of patient care and positively affects outcomes for patients and families. There is strong support for the benefits of patient engagement. What is less clear are the methods by which organizations can achieve authentic patient engagement in research and the ways a committee structure can support an institutional research engagement need beyond the individual investigators. In this report, we describe the mechanisms needed to support the implementation of a research-focused patient engagement strategy and lessons learned from the patient and family perspective.
Background Rosai–Dorfman–Destombes disease (RDD) is a rare histiocytic disorder with heterogeneous clinical manifestations and rare neurologic involvement. The existing clinical literature about neurologic RDD has yet to be critically examined. Methods We performed a four-database English-language systematic literature search for cases of RDD neurohistiocytosis, excluding secondary literature. Individual patient data for neurologic symptoms, disease sites, treatments, and responses were captured. Responses to first-line and second-line surgical interventions, post-surgical radiotherapy, and systemic therapies were analyzed. Results Among 4769 articles yielded by literature search, 154 articles were fully reviewed, containing data on 224 patients with neurologic RDD. 128 (83.1%) articles were single case reports. 149 (66.5%) patients were male, 74 (33.5%) female, with a median age of 37.6 years (range 2–79). Presenting neurologic symptoms included headache (45.1%), focal neurological deficits (32.6%), visual symptoms (32.1%), and seizures (24.6%). RDD involvement was multifocal in 32 (14.3%) cases. First-line treatment involved resection in 200 (89.6%) patients, with subsequent progression in 52 (26%), including 41 (78.8%) with unifocal disease. No difference was observed in progression-free survival comparing post-operative radiotherapy to no radiotherapy following partial resection. Chemotherapy given alone as first-line treatment led to complete or partial response in 3/7(43%) patients. Second-line treatments led to complete or partial response in 18/37(37.5%) patients. Mutational data were reported on 10 patients (4.46%). Conclusions This review highlights the limited published data about neurologic RDD, which presents with varied symptomatology and outcome. Further study is needed about its mutational landscape, and more effective therapies are needed for recurrent and refractory disease.
Background Non-infectious complications have become a major cause of morbidity and mortality in patients with Common Variable Immunodeficiency (CVID). The monitoring of patients with CVID prior to the development of non-infectious complications is not well defined. Objective Our objectives were to systematically review the current literature on the monitoring of CVID patients without non-infectious complications and to develop recommendations for such monitoring. Methods MEDLINE and EMBASE were searched from January 1 st , 2000 to March 25 th , 2021. Studies on any aspects of CVID monitoring were included. Studies that included only children, on monitoring CVID patients with existing non-infectious complications, or in the format of case reports were excluded. Results Nine studies on CVID monitoring, including 3 cohort studies, 3 experts’ opinions, 2 consensus statements and a single guideline report were identified. These studies revealed that clinical assessment and bloodwork were preformed every 6 to 12 months in asymptomatic patients. Some centers performed computerized tomography scan of the chest every 2–5 years to identify chronic lung disease, although the majority did chest imaging in accordance with clinical indications. Pulmonary function tests were done annually at most centers. Most studies did not address the role of abdominal imaging to screen for liver diseases or endoscopy to screen for gastric cancer in asymptomatic patients with uncomplicated CVID. Conclusions There is paucity of evidence-based information to guide the routine monitoring of CVID patients without non-infectious complications. Prospective studies are needed to determine the best monitoring practices in this group of patients.
Background One reason that asthma remains poorly controlled in children is poor inhaler technique. Guidelines recommend checking inhaler technique at each clinical visit. However, they do not specify how best to train children to mastery of correct inhaler technique. Many children are simply shown how to use inhalers which results in less than 50% with correct inhaler technique. The aim of this scoping review is to explore published literature on teaching methods used to train children to master correct inhaler technique. Methods We searched (from inception onwards): Medline, Embase, Scopus, Web of Science, CINAHL and the Cochrane library. We included quantitative studies, (e.g. randomised controlled trials, cohort studies and case-control studies), published from 1956 to present, on teaching inhaler technique to children with asthma. Data was extracted onto a data charting table to create a descriptive summary of the results. Data was then synthesised with descriptive statistics and visual mapping. Results Thirty-three papers were identified for full text analysis. Educational interventions were found to be taking place in a variety of clinical areas and by a range of healthcare professional disciplines. ‘Brief-Instruction’ and ‘Teach-Back’ were identified as two primary methods of providing inhaler technique training in the majority of papers. Secondary themes identified were; use of written instruction, physical demonstration, video demonstrations and/or use of inhaler devices to augment inhaler technique training. Conclusion There are a variety of means by which inhaler technique has been taught to children. These methods are likely applicable to all inhaler types and often involve some form of physical demonstration. Children of all ages can be trained to use their inhaler correctly and by a range of healthcare professionals. We have not analysed the effectiveness of these different interventions, but have described what has been trialled before in an attempt to focus our attentions on what may potentially work best. The majority of these methods can be dichotomised to either ‘Brief-Intervention’ or ‘Teach-Back’. Based on our analysis of this scoping review, we consider the following as areas for future research; how many times does a given intervention have to be done in order to have the desired effect? For what duration does the intervention need to continue to have a long-lasting effect? And, what is the best outcome measure for inhaler technique?. Trial registration Systematic review registration: Open Science Framework (
Background External chest-wall compression (ECC) is sometimes used in ARDS patients despite lack of evidence. It is currently unknown whether this practice has any clinical benefit in patients with COVID-19 ARDS (C-ARDS) characterized by a respiratory system compliance ( C rs ) < 35 mL/cmH 2 O. Objectives To test if an ECC with a 5 L-bag in low-compliance C-ARDS can lead to a reduction in driving pressure (DP) and improve gas exchange, and to understand the underlying mechanisms. Methods Eleven patients with low-compliance C-ARDS were enrolled and underwent 4 steps: baseline, ECC for 60 min, ECC discontinuation and PEEP reduction. Respiratory mechanics, gas exchange, hemodynamics and electrical impedance tomography were recorded. Four pigs with acute ARDS were studied with ECC to understand the effect of ECC on pleural pressure gradient using pleural pressure transducers in both non-dependent and dependent lung regions. Results Five minutes of ECC reduced DP from baseline 14.2 ± 1.3 to 12.3 ± 1.3 cmH 2 O ( P < 0.001), explained by an improved lung compliance. Changes in DP by ECC were strongly correlated with changes in DP obtained with PEEP reduction ( R ² = 0.82, P < 0.001). The initial benefit of ECC decreased over time (DP = 13.3 ± 1.5 cmH 2 O at 60 min, P = 0.03 vs. baseline). Gas exchange and hemodynamics were unaffected by ECC. In four pigs with lung injury, ECC led to a decrease in the pleural pressure gradient at end-inspiration [2.2 (1.1–3) vs. 3.0 (2.2–4.1) cmH 2 O, P = 0.035]. Conclusions In C-ARDS patients with C rs < 35 mL/cmH 2 O, ECC acutely reduces DP. ECC does not improve oxygenation but it can be used as a simple tool to detect hyperinflation as it improves C rs and reduces P pl gradient. ECC benefits seem to partially fade over time. ECC produces similar changes compared to PEEP reduction.
CAG-expanded ATXN7 has been previously defined in the pathogenesis of spinocerebellar ataxia type 7 (SCA7), a polyglutamine expansion autosomal dominant cerebellar ataxia. Pathology in SCA7 occurs as a result of a CAG triplet repeat expansion in excess of 37 in the first exon of ATXN7 , which encodes ataxin-7. SCA7 presents clinically with spinocerebellar ataxia and cone-rod dystrophy. Here, we present a novel spinocerebellar ataxia variant occurring in a patient with mutations in both ATXN7 and TOP1MT , which encodes mitochondrial topoisomerase I (top1mt). Using machine-guided, unbiased microscopy image analysis, we demonstrate alterations in ataxin-7 subcellular localization, and through high-fidelity measurements of cellular respiration, bioenergetic defects in association with top1mt mutations. We identify ataxin-7 Q35P and top1mt R111W as deleterious mutations, potentially contributing to disease states. We recapitulate our mutations through Drosophila genetic models. Our work provides important insight into the cellular biology of ataxin-7 and top1mt and offers insight into the pathogenesis of spinocerebellar ataxia applicable to multiple subtypes of the illness. Moreover, our study demonstrates an effective pipeline for the characterization of previously unreported genetic variants at the level of cell biology.
Congenital diaphragmatic hernia (CDH) is a rare birth defect characterized by incomplete closure of the diaphragm and herniation of fetal abdominal organs into the chest that results in pulmonary hypoplasia, postnatal pulmonary hypertension owing to vascular remodelling and cardiac dysfunction. The high mortality and morbidity rates associated with CDH are directly related to the severity of cardiopulmonary pathophysiology. Although the aetiology remains unknown, CDH has a polygenic origin in approximately one-third of cases. CDH is typically diagnosed with antenatal ultrasonography, which also aids in risk stratification, alongside fetal MRI and echocardiography. At specialized centres, prenatal management includes fetal endoscopic tracheal occlusion, which is a surgical intervention aimed at promoting lung growth in utero. Postnatal management focuses on cardiopulmonary stabilization and, in severe cases, can involve extracorporeal life support. Clinical practice guidelines continue to evolve owing to the rapidly changing landscape of therapeutic options, which include pulmonary hypertension management, ventilation strategies and surgical approaches. Survivors often have long-term, multisystem morbidities, including pulmonary dysfunction, gastroesophageal reflux, musculoskeletal deformities and neurodevelopmental impairment. Emerging research focuses on small RNA species as biomarkers of severity and regenerative medicine approaches to improve fetal lung development. Congenital diaphragmatic hernia is a rare birth defect associated with cardiopulmonary complications and high mortality and morbidity. This Primer describes the epidemiology and mechanisms of congenital diaphragmatic hernia and discusses diagnosis, management and quality of life for patients.
V-ATPases are rotary proton pumps that serve as signaling hubs with numerous protein binding partners. CryoEM with exhaustive focused classification allowed detection of endogenous proteins associated with porcine kidney V-ATPase. An extra C subunit was found in ∼3% of complexes, whereas ∼1.6% of complexes bound mEAK-7, a protein with proposed roles in dauer formation in nematodes and mTOR signaling in mammals. High-resolution cryoEM of porcine kidney V-ATPase with recombinant mEAK-7 showed that mEAK-7's TLDc domain interacts with V-ATPase's stator, whereas its C-terminal α helix binds V-ATPase's rotor. This crosslink would be expected to inhibit rotary catalysis. However, unlike the yeast TLDc protein Oxr1p, exogenous mEAK-7 does not inhibit V-ATPase and mEAK-7 overexpression in cells does not alter lysosomal or phagosomal pH. Instead, cryoEM suggests that the mEAK-7:V-ATPase interaction is disrupted by ATP-induced rotation of the rotor. Comparison of Oxr1p and mEAK-7 binding explains this difference. These results show that V-ATPase binding by TLDc domain proteins can lead to effects ranging from strong inhibition to formation of labile interactions that are sensitive to the enzyme's activity.
Background Children living with chronic comorbid conditions are at increased risk for severe COVID-19, though there is limited evidence regarding the risks associated with specific conditions and which children may benefit from targeted COVID-19 therapies. The objective of this study was to identify factors associated with severe disease among hospitalized children with COVID-19 in Canada. Methods We conducted a national prospective study on hospitalized children with microbiologically confirmed SARS-CoV-2 infection via the Canadian Paediatric Surveillance Program (CPSP) from April 2020–May 2021. Cases were reported voluntarily by a network of >2800 paediatricians. Hospitalizations were classified as COVID-19-related, incidental infection, or infection control/social admissions. Severe disease (among COVID-19-related hospitalizations only) was defined as disease requiring intensive care, ventilatory or hemodynamic support, select organ system complications, or death. Risk factors for severe disease were identified using multivariable Poisson regression, adjusting for age, sex, concomitant infections, and timing of hospitalization. Findings We identified 544 children hospitalized with SARS-CoV-2 infection, including 60·7% with COVID-19-related disease and 39·3% with incidental infection or infection control/social admissions. Among COVID-19-related hospitalizations (n=330), the median age was 1·9 years (IQR 0·1–13·3) and 43·0% had chronic comorbid conditions. Severe disease occurred in 29·7% of COVID-19-related hospitalizations (n=98/330 including 60 admitted to intensive care), most frequently among children aged 2-4 years (48·7%) and 12-17 years (41·3%). Comorbid conditions associated with severe disease included pre-existing technology dependence requirements (adjusted risk ratio [aRR] 2·01, 95% confidence interval [CI] 1·37-2·95), body mass index Z-scores ≥3 (aRR 1·90, 95% CI 1·10-3·28), neurologic conditions (e.g. epilepsy and select chromosomal/genetic conditions) (aRR 1·84, 95% CI 1·32-2·57), and pulmonary conditions (e.g. bronchopulmonary dysplasia and uncontrolled asthma) (aRR 1·63, 95% CI 1·12-2·39). Interpretation While severe outcomes were detected at all ages and among patients with and without comorbidities, neurologic and pulmonary conditions as well as technology dependence were associated with increased risk of severe COVID-19. These findings may help guide vaccination programs and prioritize targeted COVID-19 therapies for children. Funding Financial support for the CPSP was received from the Public Health Agency of Canada.
The mechanisms that link the primary increase in SR Ca2+ leak of MH susceptibility and related conditions to their disease phenotypes are not well understood. We found that abnormal Ca2+ homeostasis in MHS individuals induces proteolysis of junctophilin1 (JPh1), an essential structural protein of EC coupling (Perni, in 2017). Guo (in 2018) and Lahiri (in 2020) reported similar fragmentation of JPh2 in stressed hearts. Western blot of patients’ muscle with domain-specific antibodies showed a deficit of full-length JPh1 and excess of a 44-kD C-terminal fragment (JPh44) in MHS subjects. While JPh1 was located in T-SR junctions, JPh44 was found anywhere within the I band, and at high densities within nuclei—a location forbidden for JPh1. Expression and cleavage in mice of a JPh1 plasmid tagged at both ends showed that its N-terminal fragment remained in triads, and the C-terminal fragment, orthologue to JPh44, entered nuclei, which indicates that JPh44 is the C-terminal cleavage product. Endogenous calpain1 appeared in T-SR junctions, colocalized with JPh1. On muscle extracts and primary cultures, Ca2+-activated calpain1 cleaved a 44-kD JPh1 piece, consistent with the C-terminal fragment that starts at Ser241, the highest probability cleavage site found by calpain1 algorithms. Completing the identification of Ser241 as the likely start of JPh44, the tagged deletion plasmid GFP-JPh1_Δ1-240, expressed in mice, copied the location and migration of JPh44. Expression of GFP-JPh1_Δ1-240 in C2C12 myoblasts reduced by more than twofold the transcription of PI3K-Akt genes that inhibit muscle uptake and storage of glucose, including GSK3β, an inhibitor of glycogen synthase that is activated in MHS patients. In agreement with the genetic profile, GSK3β protein content decreased upon expression of GFP-JPh1_Δ1-240. In sum, the identified gene control roles of JPh44 oppose the deleterious effects of chronically elevated cytosolic [Ca2+], including late-onset hyperglycemia and type-2 diabetes (Tammineni, in 2020).
Study objective This study was aimed to assess the association between the use of epidural analgesia during labor and mother-infant bonding. Design A cross-sectional study. Setting Maternity ward at Soroka University Medical Center during 2020. Patients Women who delivered a singleton live-born infant vaginally in their immediate post-partum period. Interventions Women completed questionnaires. 25 items post-partum bonding questionnaire (PBQ) to assess mother-infant bonding (A high score on the PBQ indicates impaired mother-infant bonding) and the Edinburgh postnatal depression scale (EPDS) questionnaire to assess risk for post-partum depression. Measurements The study used PBQ questionnaire and four sub-scales to assess mother-infant bonding and the EPDS questionnaire to assess risk for post- partum depression. Generalized linear regression models (gamma) were constructed to examine the association between epidural analgesia and mother-infant bonding total score and impaired bonding sub- scale, while adjusting for confounders Additional information such as pregnancy complications and sociodemographic data was drawn from women's medical records. Main results A total of 234 women were included in the final analysis, of them 126 (53.8%) delivered with epidural analgesia. The total PBQ score was significantly lower among women who received epidural analgesia compared to women without epidural analgesia (7.6 vs. 10.2, p = 0.024), demonstrating a better mother -infant bonding. Using two multivariable linear regression models, controlling for confounders such as maternal age and educational status, epidural analgesia during labor was independently associated with a better mother -infant bonding total score and better impaired bonding sub-scale score (Beta coefficient-0.252, 95% CI -0.5; −0.006, p = 0.045 and Beta coefficient − 0.34, 95% CI -0.52; −0.08, p = 0.01 for mother-infant bonding total score and sub-scale score, respectively). No differences in post-partum depression risks were found between the groups (EDPS≥13, 5.7% vs. 13%, p = 0.058). Conclusion Our study demonstrated better mother -infant bonding among women delivering with epidural analgesia.
Background and Objectives Machine Learning offers opportunities to improve patient outcomes, team performance, and reduce healthcare costs. Yet only a small fraction of all Machine Learning models for health care have been successfully integrated into the clinical space. There are no current guidelines for clinical model integration, leading to waste, unnecessary costs, patient harm, and decreases in efficiency when improperly implemented. Systems engineering is widely used in industry to achieve an integrated system of systems through an interprofessional collaborative approach to system design, development, and integration. We propose a framework based on systems engineering to guide the development and integration of Machine Learning models in healthcare. Methods Applied systems engineering, software engineering and health care Machine Learning software development practices were reviewed and critically appraised to establish an understanding of limitations and challenges within these domains. Principles of systems engineering were used to develop solutions to address the identified problems. The framework was then harmonized with the Machine Learning software development process to create a systems engineering-based Machine Learning software development approach in the healthcare domain. Results We present an integration framework for healthcare Artificial Intelligence that considers the entirety of this system of systems . Our proposed framework utilizes a combined software and integration engineering approach and consists of four phases: (1) Inception, (2) Preparation, (3) Development, and (4) Integration. During each phase, we present specific elements for consideration in each of the three domains of integration: The Human, The Technical System, and The Environment. There are also elements that are considered in the interactions between these domains. Conclusion Clinical models are technical systems that need to be integrated into the existing system of systems in health care. A systems engineering approach to integration ensures appropriate elements are considered at each stage of model design to facilitate model integration. Our proposed framework is based on principles of systems engineering and can serve as a guide for model development, increasing the likelihood of successful Machine Learning translation and integration.
Institution pages aggregate content on ResearchGate related to an institution. The members listed on this page have self-identified as being affiliated with this institution. Publications listed on this page were identified by our algorithms as relating to this institution. This page was not created or approved by the institution. If you represent an institution and have questions about these pages or wish to report inaccurate content, you can contact us here.
1,989 members
Clyde Matava
  • Department of Anesthesia and Pain Medicine
Aleksander Hinek
  • Translational Medicine
Robert Murray Hamilton
  • Labatt Family Heart Centre
Anirban Das
  • Division of Hematology/Oncology
555 University Avenue, M5G 1X8, Toronto, Ontario, Canada
Head of institution
Ronald Cohn