Wei, Wei‐Qi

Background: Susceptibility to cardiometabolic diseases (CDs) is shaped by genetic, socio-environmental, and lifestyle factors. We evaluated the associations of disease-specific polysocial risk scores (PSSs)— derived from social determinants of health (SDOH) and lifestyle factors—and genetic risk factors with six CDs, across self-identified race/ethnicity (SIRE) groups in the eMERGE IV cohort. Methods: The six CDs included: atrial fibrillation, coronary heart disease, chronic kidney disease, hyperlipidemia, obesity, and type 2 diabetes. Disease-specific PSSs were developed using machine-learning models based on 65 SDOH and lifestyle factors. Participants were grouped based on their SIRE as Asian, Black, Hispanic/Latino, and White. We evaluated PSS distributions across SIRE groups. Genetic risk was assessed through disease-specific polygenic risk scores (PRSs), monogenic etiologies, and family history (FamHx). High PSS and high PRS were defined as the top 5th percentile. Associations were tested using multivariable logistic regression; model performance was assessed using the C-statistic. Results: Among 20178 participants (50±15 years, 68% female, 40% non-White), the most frequently associated PSS components were physical activity, sleep, self-perceived health, and income. PSS distributions varied across SIRE groups, with Whites having the lowest and Black and Hispanic/Latinos the highest scores for most CDs (Figure 1). All PSSs were significantly associated with their respective CDs, independent of confounders (OR per 1 SD increase: 1.19–2.68; Figure 2-A). No significant interactions were observed between PSSs and PRSs or monogenic etiologies. However, significant interactions were noted between PSSs and FamHx for CKD and obesity. PSS and PRS effects on the odds of CDs were independent (Figure 2-B) and additive (Figure 2-C). Incorporating PSSs into models that included clinical and genetic risk factors improved predictive performance of risk prediction models (Figure 3). Conclusion: PSSs, reflecting socio-environmental and lifestyle disadvantage, varied significantly across SIRE groups and were associated with risk of six CDs, independent of and additive to PRS. Given the substantial variation in socio-environmental and lifestyle risks across SIRE groups, integrating these factors into risk prediction models for CDs may help address health disparities and enhance preventive efforts, especially in marginalized populations.

Background: We investigated the impact of genetic risk factors for coronary heart disease (CHD) -polygenic risk score (PRS), familial hypercholesterolemia (FH), and family history (FamHx)- on CHD risk estimates, across the age spectrum, in two diverse cohorts of US adults -eMERGE IV (eIV) and All of Us (AoU). Methods: CHD was defined as myocardial infarction, unstable angina, and coronary revascularization. Self-identified race/ethnicity (SIRE) was used as a population descriptor. We calculated a PRS for CHD (PRS CHD , PGS004698), ascertained FH as the presence of pathogenic/likely pathogenic variants in FH genes, and defined FamHx as early-onset CHD in first-degree relatives. We employed Pooled Cohort Equations (PCE) to estimate the 10y risk of CHD for adults ≥40y and modeled the association of conventional risk factors with CHD in adults <40y. AoU served as the training set, and eIV as the testing set. We analyzed the impact of PRS CHD and FamHx on CHD risk estimates by a) using multivariable logistic regression and Cox proportional hazard models, assessing discrimination and the extent of risk reclassification; and b) net benefit analysis and decision curves to assess the performance of prediction models across actionable thresholds. Results: We analyzed data for 19348 participants from eIV (age 50±15, 68% female, 41% non-White) and 239645 participants from AoU (age 55±17, 61% female, 48% non-White). Genetic risk factors were significantly associated with CHD. PRS CHD performance varied by SIRE groups, while FamHx was consistent. The effects of PRS CHD and FamHx on CHD were independent and additive (Figure 1). In adults ≥40y, incorporating PRS CHD and FamHx into PCE improved discrimination (C-statistic increased from 0.719 to 0.753; P -diff=9.1×10 -3 , Figure 2) and reclassified risk in 19% and 20% of participants at the 7.5% and 10% 10y CHD risk thresholds, respectively. Between the 7.5% and 10% 10y CHD risk thresholds, incorporating PRS CHD and FamHx into the PCE improved the net benefit of the risk prediction models across White, Black, and Hispanic/Latino groups (Figure 3). Conclusion: PRS CHD and FamHx were independently and additively associated with CHD in two large diverse cohorts in the US. Incorporating PRS CHD and FamHx into PCE improved risk discrimination, reclassified risk in a significant portion of participants, and improved net benefit of the PCE across all three major SIRE groups, motivating the addition of these factors to clinical risk calculators.

Abstract Objectives To implementation an automated multi-institutional pipeline that delivers breast-cancer risk integrated with polygenic risk scores, monogenic variants, family history, and clinical factors, emphasizing operational challenges and their solutions. Materials and Methods A five-stage process was executed at ten sites. Data streams from REDCap surveys, PRS and monogenic reports, and MeTree pedigrees were normalized and forwarded through a REDCap plug-in to the CanRisk API. Results Integrated risk was returned to &gt;10 000 women; 3.6% were ≥25 % lifetime risk and 0.9% carried pathogenic variants. Pipeline generated score aligns well with manual generated ones. Major barriers such as heterogeneous pedigree formats, missing data, edge-case handling, and evolving model versions were identified and resolved through mapping rules, imputations, and iterative testing. Discussion Cross-platform data harmonization and stakeholder alignment were decisive for success. Borderline-risk communication and model-version drift remain open issues. Conclusion Large-scale PRS-integrated breast-cancer risk reporting is feasible but requires robust interoperability standards and iterative governance.

Apart from ancestry, personal or environmental covariates may contribute to differences in polygenic score (PGS) performance. We analyzed the effects of covariate stratification and interaction on body mass index (BMI) PGS (PGS BMI ) across four cohorts of European (N = 491,111) and African (N = 21,612) ancestry. Stratifying on binary covariates and quintiles for continuous covariates, 18/62 covariates had significant and replicable R 2 differences among strata. Covariates with the largest differences included age, sex, blood lipids, physical activity, and alcohol consumption, with R 2 being nearly double between best- and worst-performing quintiles for certain covariates. Twenty-eight covariates had significant PGS BMI –covariate interaction effects, modifying PGS BMI effects by nearly 20% per standard deviation change. We observed overlap between covariates that had significant R 2 differences among strata and interaction effects – across all covariates, their main effects on BMI were correlated with their maximum R 2 differences and interaction effects (0.56 and 0.58, respectively), suggesting high-PGS BMI individuals have highest R 2 and increase in PGS effect. Using quantile regression, we show the effect of PGS BMI increases as BMI itself increases, and that these differences in effects are directly related to differences in R 2 when stratifying by different covariates. Given significant and replicable evidence for context-specific PGS BMI performance and effects, we investigated ways to increase model performance taking into account nonlinear effects. Machine learning models (neural networks) increased relative model R 2 (mean 23%) across datasets. Finally, creating PGS BMI directly from GxAge genome-wide association studies effects increased relative R 2 by 7.8%. These results demonstrate that certain covariates, especially those most associated with BMI, significantly affect both PGS BMI performance and effects across diverse cohorts and ancestries, and we provide avenues to improve model performance that consider these effects.

Abstract Current studies regarding the secondary use of electronic health records (EHR) predominantly rely on domain expertise and existing medical knowledge. Though significant efforts have been devoted to investigating the application of machine learning algorithms in the EHR, efficient and powerful representation of patients is needed to unleash the potential of discovering new medical patterns underlying the EHR. Here, we present an unsupervised method for embedding high-dimensional EHR data at the patient level, aimed at characterizing patient heterogeneity in complex diseases and identifying new disease patterns associated with clinical outcome disparities. Inspired by the architecture of modern language models—specifically transformers with attention mechanisms, we use patient diagnosis and procedure codes as vocabularies and treat each patient as a sentence to perform the patient embedding. We applied this approach to 34,851 unique medical codes across 1,046,649 longitudinal patient events, including 102,739 patients from the electronic Medical Records and GEnomics (eMERGE) Network. The resulting patient vectors demonstrated excellent performance in predicting future disease events (median AUROC = 0.87 within one year) and bulk phenotyping (median AUROC = 0.84). We then illustrated the utility of these patient vectors in revealing heterogeneous comorbidity patterns, exemplified by disease subtypes in colorectal cancer and systemic lupus erythematosus, and capturing distinct longitudinal disease trajectories. External validation using EHR data from the University of Washington confirmed robust model performance, with median AUROCs of 0.83 and 0.84 for bulk phenotyping tasks and disease onset prediction, respectively. Importantly, the model reproduced the clustering results of disease subtypes identified in the eMERGE cohort and uncovered variations in overall mortality among these subtypes. Together, these results underscore the potential of representation learning in EHRs to enhance patient characterization and associated clinical outcomes, thereby advancing disease forecasting and facilitating personalized medicine.

Apart from ancestry, personal or environmental covariates may contribute to differences in polygenic score (PGS) performance. We analyzed the effects of covariate stratification and interaction on body mass index (BMI) PGS (PGS BMI ) across four cohorts of European (N = 491,111) and African (N = 21,612) ancestry. Stratifying on binary covariates and quintiles for continuous covariates, 18/62 covariates had significant and replicable R 2 differences among strata. Covariates with the largest differences included age, sex, blood lipids, physical activity, and alcohol consumption, with R 2 being nearly double between best- and worst-performing quintiles for certain covariates. Twenty-eight covariates had significant PGS BMI –covariate interaction effects, modifying PGS BMI effects by nearly 20% per standard deviation change. We observed overlap between covariates that had significant R 2 differences among strata and interaction effects – across all covariates, their main effects on BMI were correlated with their maximum R 2 differences and interaction effects (0.56 and 0.58, respectively), suggesting high-PGS BMI individuals have highest R 2 and increase in PGS effect. Using quantile regression, we show the effect of PGS BMI increases as BMI itself increases, and that these differences in effects are directly related to differences in R 2 when stratifying by different covariates. Given significant and replicable evidence for context-specific PGS BMI performance and effects, we investigated ways to increase model performance taking into account nonlinear effects. Machine learning models (neural networks) increased relative model R 2 (mean 23%) across datasets. Finally, creating PGS BMI directly from GxAge genome-wide association studies effects increased relative R 2 by 7.8%. These results demonstrate that certain covariates, especially those most associated with BMI, significantly affect both PGS BMI performance and effects across diverse cohorts and ancestries, and we provide avenues to improve model performance that consider these effects.

AbstractObjectiveThe aim of this study was to analyze a publicly available sample of rule-based phenotype definitions to characterize and evaluate the variability of logical constructs used.Materials and MethodsA sample of 33 preexisting phenotype definitions used in research that are represented using Fast Healthcare Interoperability Resources and Clinical Quality Language (CQL) was analyzed using automated analysis of the computable representation of the CQL libraries.ResultsMost of the phenotype definitions include narrative descriptions and flowcharts, while few provide pseudocode or executable artifacts. Most use 4 or fewer medical terminologies. The number of codes used ranges from 5 to 6865, and value sets from 1 to 19. We found that the most common expressions used were literal, data, and logical expressions. Aggregate and arithmetic expressions are the least common. Expression depth ranges from 4 to 27.DiscussionDespite the range of conditions, we found that all of the phenotype definitions consisted of logical criteria, representing both clinical and operational logic, and tabular data, consisting of codes from standard terminologies and keywords for natural language processing. The total number and variety of expressions are low, which may be to simplify implementation, or authors may limit complexity due to data availability constraints.ConclusionsThe phenotype definitions analyzed show significant variation in specific logical, arithmetic, and other operators but are all composed of the same high-level components, namely tabular data and logical expressions. A standard representation for phenotype definitions should support these formats and be modular to support localization and shared logic.

AbstractObjectiveHigh BMI is associated with many comorbidities and mortality. This study aimed to elucidate the overall clinical risk of obesity using a genome‐ and phenome‐wide approach.MethodsThis study performed a phenome‐wide association study of BMI using a clinical cohort of 736,726 adults. This was followed by genetic association studies using two separate cohorts: one consisting of 65,174 adults in the Electronic Medical Records and Genomics (eMERGE) Network and another with 405,432 participants in the UK Biobank.ResultsClass 3 obesity was associated with 433 phenotypes, representing 59.3% of all billing codes in individuals with severe obesity. A genome‐wide polygenic risk score for BMI, accounting for 7.5% of variance in BMI, was associated with 296 clinical diseases, including strong associations with type 2 diabetes, sleep apnea, hypertension, and chronic liver disease. In all three cohorts, 199 phenotypes were associated with class 3 obesity and polygenic risk for obesity, including novel associations such as increased risk of renal failure, venous insufficiency, and gastroesophageal reflux.ConclusionsThis combined genomic and phenomic systematic approach demonstrated that obesity has a strong genetic predisposition and is associated with a considerable burden of disease across all disease classes.