A5070430680- Machine Learning in Healthcare
- Diabetes and associated disorders
- Diabetes Management and Research
- Pancreatic function and diabetes
- Genetic Associations and Epidemiology
- Artificial Intelligence in Healthcare
- Cardiovascular Function and Risk Factors
- Speech Recognition and Synthesis
- Lipoproteins and Cardiovascular Health
- Health Systems, Economic Evaluations, Quality of Life
- Advanced Causal Inference Techniques
- Liver Disease Diagnosis and Treatment
- Diabetes, Cardiovascular Risks, and Lipoproteins
- Statistical Methods and Inference
- Cardiovascular Disease and Adiposity
- Chronic Disease Management Strategies
- Speech and dialogue systems
- Topic Modeling
- Natural Language Processing Techniques
- Cardiovascular Health and Risk Factors
- Neural Networks and Applications
- Artificial Intelligence in Healthcare and Education
- Cardiomyopathy and Myosin Studies
- Speech and Audio Processing
- Data Visualization and Analytics
IBM (United States)
2015-2024
Computational Physics (United States)
2018-2024
Eli Lilly (United States)
2024
Cambridge Scientific (United States)
2017-2024
Massachusetts Institute of Technology
2000-2024
University of Oulu
2021-2023
Helmholtz Zentrum München
2021-2023
University of Colorado Anschutz Medical Campus
2021-2023
Lund University
2021-2023
Pacific Northwest Diabetes Research Institute
2023
Understanding predictive models, in terms of interpreting and identifying actionable insights, is a challenging task. Often the importance feature model only rough estimate condensed into one number. However, our research goes beyond these naïve estimates through design implementation an interactive visual analytics system, Prospector. By providing partial dependence diagnostics, data scientists can understand how features affect prediction overall. In addition, support for localized...
Abstract Genetic variation can predispose to disease both through (i) monogenic risk variants that disrupt a physiologic pathway with large effect on and (ii) polygenic involves many of small in different pathways. Few studies have explored the interplay between risk. Here, we study 80,928 individuals examine whether background modify penetrance tier 1 genomic conditions — familial hypercholesterolemia, hereditary breast ovarian cancer, Lynch syndrome. Among carriers variant, estimate...
Objective: Lp(a) (lipoprotein[a]) concentrations are associated with atherosclerotic cardiovascular disease (ASCVD), and new therapies that enable potent specific reduction in development. In the largest study conducted to date, we address 3 areas of uncertainty: (1) magnitude shape ASCVD risk conferred across distribution lipoprotein(a) concentrations; (2) variation racial clinical subgroups; (3) importance a high threshold guide therapy. Approach Results: Relationship incident was studied...
Background: Individuals of South Asian ancestry represent 23% the global population, corresponding to 1.8 billion people, and have substantially higher risk atherosclerotic cardiovascular disease compared with most other ethnicities. US practice guidelines now recognize as an important risk-enhancing factor. The magnitude enhanced within context contemporary clinical care, extent which it is captured by existing estimators, its potential mechanisms warrant additional study. Methods: Within...
For any given level of overall adiposity, individuals vary considerably in fat distribution. The inherited basis distribution the general population is not fully understood. Here, we study up to 38,965 UK Biobank participants with MRI-derived visceral (VAT), abdominal subcutaneous (ASAT), and gluteofemoral (GFAT) adipose tissue volumes. Because these depot volumes are highly correlated BMI, additionally six local adiposity traits: VAT adjusted for BMI height (VATadj), ASATadj, GFATadj,...
For any given body mass index (BMI), individuals vary substantially in fat distribution, and this variation may have important implications for cardiometabolic risk. Here, we study disease associations with BMI-independent visceral (VAT), abdominal subcutaneous (ASAT), gluteofemoral (GFAT) depots 40,032 of the UK Biobank MRI. We apply deep learning models based on two-dimensional MRI projections to enable near-perfect estimation depot volumes (R
Clustering, the process of grouping together similar items into distinct partitions, is a common type unsupervised machine learning that can be useful for summarizing and aggregating complex multi-dimensional data. However, data clustered in many ways, there exist large body algorithms designed to reveal different patterns. While having access wide variety helpful, practice, it quite difficult scientists choose parameterize get clustering results relevant their dataset analytical tasks. To...
To determine the relationship of a genome-wide polygenic score for coronary artery disease (GPSCAD) with lifetime trajectories CAD risk, directly compare its predictive capacity to traditional risk factors, and assess interplay Pooled Cohort Equations (PCE) clinical estimator. Approach Results: We studied GPSCAD in 28 556 middle-aged participants Malmö Diet Cancer Study, whom 4122 (14.4%) developed over median follow-up 21.3 years. A pronounced gradient was observed-16% those lowest decile...
Background— Using electronic health records data to predict events and onset of diseases is increasingly common. Relatively little known, although, about the tradeoffs between requirements model utility. Methods Results— We examined performance machine learning models trained detect prediagnostic heart failure in primary care patients using longitudinal data. Model was assessed relation defined by prediction window length (time before clinical diagnosis), observation (duration window),...
<h3>Importance</h3> Pathogenic DNA variants associated with familial hypercholesterolemia, hereditary breast and ovarian cancer syndrome, Lynch syndrome are widely recognized as clinically important actionable when identified, leading some clinicians to recommend population-wide genomic screening. <h3>Objectives</h3> To assess the prevalence clinical importance of pathogenic or likely each 3 conditions (familial syndrome) within context contemporary care. <h3>Design, Setting,...
BackgroundParkinson's disease is heterogeneous in symptom presentation and progression. Increased understanding of both aspects can enable better patient management improve clinical trial design. Previous approaches to modelling Parkinson's progression assumed static trajectories within subgroups have not adequately accounted for complex medication effects. Our objective was develop a statistical model that accounts intra-individual inter-individual variability effects.MethodsIn this...
A fundamental challenge in diagnostics is integrating multiple modalities to develop a joint characterization of physiological state. Using the heart as model system, we cross-modal autoencoder framework for distinct data and constructing holistic representation cardiovascular In particular, use our construct such representations from cardiac magnetic resonance images (MRIs), containing structural information, electrocardiograms (ECGs), myoelectric information. We leverage learned (1)...
Abstract Increased left atrial volume and decreased function have long been associated with fibrillation. The availability of large-scale cardiac magnetic resonance imaging data paired genetic provides a unique opportunity to assess the contributions structure function, understand their relationship risk for Here, we use deep learning surface reconstruction models measure minimum volume, maximum stroke emptying fraction in 40,558 UK Biobank participants. In genome-wide association study...
Background and aims Accurate assessment of the risk mortality following a cirrhosis-related admission can enable health-care providers to identify high-risk patients modify treatment plans decrease mortality. Methods We developed post-discharge prediction model for with using population 314,292 who received care either at Massachusetts General Hospital (MGH) or Brigham Women's (BWH) between 1992 2010. extracted 68 variables from electronic medical records (EMRs), including demographics,...
Clinical researchers use disease progression models to understand patient status and characterize patterns from longitudinal health records. One approach for modeling is describe using a small number of states that represent distinctive distributions over set observed measures. Hidden Markov (HMMs) its variants are class both discover these make inferences patients. Despite the advantages algorithms discovering interesting patterns, it still remains challenging medical experts interpret...
OBJECTIVE To combine prospective cohort studies, by including HLA harmonization, and estimate risk of islet autoimmunity progression to clinical diabetes. RESEARCH DESIGN AND METHODS For cohorts in Finland, Germany, Sweden, the U.S., 24,662 children at increased genetic for development autoantibodies type 1 diabetes have been followed. Following outcomes were analyzed 16,709 infants-toddlers enrolled age 2.5 years. RESULTS In infant-toddler cohort, 1,413 (8.5%) developed least one...
<h3>Importance</h3> Familial hypercholesterolemia variants impair clearance of cholesterol from the circulation and increase risk coronary artery disease (CAD). The extent to which adherence a healthy lifestyle is associated with lower CAD in carriers noncarriers warrants further study. <h3>Objective</h3> To assess association interaction between familial CAD. <h3>Design, Setting, Participants</h3> This cross-sectional study used 2 independent data sets gene sequencing UK Biobank:...