Characterizing genetic profiles for high triglyceride levels in U.S. patients of African ancestry.

Hypertriglyceridemia (HTG) is a common cardiovascular risk factor characterized by elevated triglyceride (TG) levels. Researchers have assessed the genetic factors that influence HTG in studies focused predominantly on individuals of European ancestry. However, relatively little is known about the contribution of genetic variation of HTG in people of African ancestry (AA), potentially constraining research and treatment opportunities. Our objective was to characterize genetic profiles among individuals of AA with mild-to-moderate HTG and severe HTG versus those with normal TGs by leveraging whole-genome sequencing data and longitudinal electronic health records available in the All of Us program. We compared the enrichment of functional variants within five canonical TG metabolism genes, an AA-specific polygenic risk score for TGs, and frequencies of 145 known potentially causal TG variants between HTG patients and normal TG among a cohort of AA patients (N = 15,373). Those with mild-to-moderate HTG (N = 342) and severe HTG (N ≤ 20) were more likely to carry APOA5 p.S19W (odds ratio = 1.94, 95% confidence interval = [1.48-2.54], P = 1.63 × 10-6 and OR = 3.65, 95% confidence interval: [1.22-10.93], P = 0.02, respectively) than those with normal TG. They were also more likely to have an elevated (top 10%) polygenic risk score, elevated carriage of potentially causal variant alleles, and carry any genetic risk factor. Alternative definitions of HTG yielded comparable results. In conclusion, individuals of AA with HTG were enriched for genetic risk factors compared to individuals with normal TGs.

Deep learning for virtual orthodontic bracket removal: tool establishment and application.

OBJECTIVE: We aimed to develop a tool for virtual orthodontic bracket removal based on deep learning algorithms for feature extraction from bonded teeth and to demonstrate its application in a bracket position assessment scenario. MATERIALS AND METHODS: Our segmentation network for virtual bracket removal was trained using dataset A, containing 978 bonded teeth, 20 original teeth, and 20 brackets generated by scanners. The accuracy and segmentation time of the network were tested by dataset B, which included an additional 118 bonded teeth without knowing the original tooth morphology. This tool was then applied for bracket position assessment. The clinical crown center, bracket center, and orientations of separated teeth and brackets were extracted for analyzing the linear distribution and angular deviation of bonded brackets. RESULTS: This tool performed virtual bracket removal in 2.9 ms per tooth with accuracies of 98.93% and 97.42% (P < 0.01) in datasets A and B, respectively. The tooth surface and bracket characteristics were extracted and used to evaluate the results of manually bonded brackets by 49 orthodontists. Personal preferences for bracket angulation and bracket distribution were displayed graphically and tabularly. CONCLUSIONS: The tool's efficiency and precision are satisfactory, and it can be operated without original tooth data. It can be used to display the bonding deviation in the bracket position assessment scenario. CLINICAL SIGNIFICANCE: With the aid of this tool, unnecessary bracket removal can be avoided when evaluating bracket positions and modifying treatment plans. It has the potential to produce retainers and orthodontic devices prior to tooth debonding.

Individualized Dose-Response to Statins Associated with Cardiovascular Disease Outcomes.

Background: Statins reduce low-density lipoprotein cholesterol (LDL-C) and are efficacious in the prevention of atherosclerotic cardiovascular disease (ASCVD). Dose-response to statins varies among patients and can be modeled using three distinct pharmacological properties: (1) E0 (baseline LDL-C), (2) ED50 (potency: median dose achieving 50% reduction in LDL-C); and (3) Emax (efficacy: maximum LDL-C reduction). However, individualized dose-response and its association with ASCVD events remains unknown. Objective: We analyze the relationship between ED50 and Emax with real-world cardiovascular disease outcomes. Method: We leveraged de-identified electronic health record data to identify individuals exposed to multiple doses of the three most commonly prescribed statins (atorvastatin, simvastatin, or rosuvastatin) within the context of their longitudinal healthcare. We derived ED50 and Emax to quantify the relationship with a composite outcome of ASCVD events and all-cause mortality. Results: We estimated ED50 and Emax for 3,033 unique individuals (atorvastatin: 1,632, simvastatin: 1,089, and rosuvastatin: 312) using a nonlinear, mixed effects dose-response model. Time-to-event analyses revealed that ED50 and Emax are independently associated with the primary endpoint. Hazard ratios were 0.85 (p < 0.01), 0.83 (p < 0.01), and 0.87 (p = 0.10) for ED50 and 1.13 (p < 0.001), 1.06 (p < 0.001), and 1.15 (p = 0.009) for Emax in the atorvastatin, simvastatin, and rosuvastatin cohorts, respectively. Conclusion: The class-wide association of ED50 and Emax with clinical outcomes indicates that these measures influence the risk for ASCVD events in patients on statins.

Pharmacogenomics polygenic risk score: Ready or not for prime time?

Pharmacogenomic Polygenic Risk Scores (PRS) have emerged as a tool to address the polygenic nature of pharmacogenetic phenotypes, increasing the potential to predict drug response. Most pharmacogenomic PRS have been extrapolated from disease-associated variants identified by genome wide association studies (GWAS), although some have begun to utilize genetic variants from pharmacogenomic GWAS. As pharmacogenomic PRS hold the promise of enabling precision medicine, including stratified treatment approaches, it is important to assess the opportunities and challenges presented by the current data. This assessment will help determine how pharmacogenomic PRS can be advanced and transitioned into clinical use. In this review, we present a summary of recent evidence, evaluate the current status, and identify several challenges that have impeded the progress of pharmacogenomic PRS. These challenges include the reliance on extrapolations from disease genetics and limitations inherent to pharmacogenomics research such as low sample sizes, phenotyping inconsistencies, among others. We finally propose recommendations to overcome the challenges and facilitate the clinical implementation. These recommendations include standardizing methodologies for phenotyping, enhancing collaborative efforts, developing new statistical methods to capitalize on drug-specific genetic associations for PRS construction. Additional recommendations include enhancing the infrastructure that can integrate genomic data with clinical predictors, along with implementing user-friendly clinical decision tools, and patient education. Ethical and regulatory considerations should address issues related to patient privacy, informed consent and safe use of PRS. Despite these challenges, ongoing research and large-scale collaboration is likely to advance the field and realize the potential of pharmacogenomic PRS.

Leveraging generative AI to prioritize drug repurposing candidates for Alzheimer's disease with real-world clinical validation.

Drug repurposing represents an attractive alternative to the costly and time-consuming process of new drug development, particularly for serious, widespread conditions with limited effective treatments, such as Alzheimer's disease (AD). Emerging generative artificial intelligence (GAI) technologies like ChatGPT offer the promise of expediting the review and summary of scientific knowledge. To examine the feasibility of using GAI for identifying drug repurposing candidates, we iteratively tasked ChatGPT with proposing the twenty most promising drugs for repurposing in AD, and tested the top ten for risk of incident AD in exposed and unexposed individuals over age 65 in two large clinical datasets: (1) Vanderbilt University Medical Center and (2) the All of Us Research Program. Among the candidates suggested by ChatGPT, metformin, simvastatin, and losartan were associated with lower AD risk in meta-analysis. These findings suggest GAI technologies can assimilate scientific insights from an extensive Internet-based search space, helping to prioritize drug repurposing candidates and facilitate the treatment of diseases.

Large Language Models Facilitate the Generation of Electronic Health Record Phenotyping Algorithms.

Objectives: Phenotyping is a core task in observational health research utilizing electronic health records (EHRs). Developing an accurate algorithm demands substantial input from domain experts, involving extensive literature review and evidence synthesis. This burdensome process limits scalability and delays knowledge discovery. We investigate the potential for leveraging large language models (LLMs) to enhance the efficiency of EHR phenotyping by generating high-quality algorithm drafts. Materials and Methods: We prompted four LLMs-GPT-4 and GPT-3.5 of ChatGPT, Claude 2, and Bard-in October 2023, asking them to generate executable phenotyping algorithms in the form of SQL queries adhering to a common data model (CDM) for three phenotypes (i.e., type 2 diabetes mellitus, dementia, and hypothyroidism). Three phenotyping experts evaluated the returned algorithms across several critical metrics. We further implemented the top-rated algorithms and compared them against clinician-validated phenotyping algorithms from the Electronic Medical Records and Genomics (eMERGE) network. Results: GPT-4 and GPT-3.5 exhibited significantly higher overall expert evaluation scores in instruction following, algorithmic logic, and SQL executability, when compared to Claude 2 and Bard. Although GPT-4 and GPT-3.5 effectively identified relevant clinical concepts, they exhibited immature capability in organizing phenotyping criteria with the proper logic, leading to phenotyping algorithms that were either excessively restrictive (with low recall) or overly broad (with low positive predictive values). Conclusion: GPT versions 3.5 and 4 are capable of drafting phenotyping algorithms by identifying relevant clinical criteria aligned with a CDM. However, expertise in informatics and clinical experience is still required to assess and further refine generated algorithms.

Characterizing genetic profiles for high triglyceride levels in U.S. patients of African ancestry.

Hypertriglyceridemia (HTG) is a common cardiovascular risk factor characterized by elevated circulating triglyceride (TG) levels. Researchers have assessed the genetic factors that influence HTG in studies focused predominantly on individuals of European ancestry (EA). However, relatively little is known about the contribution of genetic variation to HTG in people of AA, potentially constraining research and treatment opportunities; the lipid profile for African ancestry (AA) populations differs from that of EA populations-which may be partially attributable to genetics. Our objective was to characterize genetic profiles among individuals of AA with mild-to-moderate HTG and severe HTG versus those with normal TGs by leveraging whole genome sequencing (WGS) data and longitudinal electronic health records (EHRs) available in the All of Us (AoU) program. We compared the enrichment of functional variants within five canonical TG metabolism genes, an AA-specific polygenic risk score for TGs, and frequencies of 145 known potentially causal TG variants between patients with HTG and normal TG among a cohort of AA patients (N=15,373). Those with mild-to-moderate HTG (N=342) and severe HTG (N≤20) were more likely to carry APOA5 p.S19W (OR=1.94, 95% CI [1.48-2.54], p=1.63×10 -6 and OR=3.65, 95% CI [1.22-10.93], p=0.02, respectively) than those with normal TG. They were also more likely to have an elevated (top 10%) PRS, elevated carriage of potentially causal variant alleles, and carry any genetic risk factor. Alternative definitions of HTG yielded comparable results. In conclusion, individuals of AA with HTG were enriched for genetic risk factors compared to individuals with normal TGs.

Large language models facilitate the generation of electronic health record phenotyping algorithms.

OBJECTIVES: Phenotyping is a core task in observational health research utilizing electronic health records (EHRs). Developing an accurate algorithm demands substantial input from domain experts, involving extensive literature review and evidence synthesis. This burdensome process limits scalability and delays knowledge discovery. We investigate the potential for leveraging large language models (LLMs) to enhance the efficiency of EHR phenotyping by generating high-quality algorithm drafts. MATERIALS AND METHODS: We prompted four LLMs-GPT-4 and GPT-3.5 of ChatGPT, Claude 2, and Bard-in October 2023, asking them to generate executable phenotyping algorithms in the form of SQL queries adhering to a common data model (CDM) for three phenotypes (ie, type 2 diabetes mellitus, dementia, and hypothyroidism). Three phenotyping experts evaluated the returned algorithms across several critical metrics. We further implemented the top-rated algorithms and compared them against clinician-validated phenotyping algorithms from the Electronic Medical Records and Genomics (eMERGE) network. RESULTS: GPT-4 and GPT-3.5 exhibited significantly higher overall expert evaluation scores in instruction following, algorithmic logic, and SQL executability, when compared to Claude 2 and Bard. Although GPT-4 and GPT-3.5 effectively identified relevant clinical concepts, they exhibited immature capability in organizing phenotyping criteria with the proper logic, leading to phenotyping algorithms that were either excessively restrictive (with low recall) or overly broad (with low positive predictive values). CONCLUSION: GPT versions 3.5 and 4 are capable of drafting phenotyping algorithms by identifying relevant clinical criteria aligned with a CDM. However, expertise in informatics and clinical experience is still required to assess and further refine generated algorithms.

Risk factors affecting polygenic score performance across diverse cohorts.

Apart from ancestry, personal or environmental covariates may contribute to differences in polygenic score (PGS) performance. We analyzed effects of covariate stratification and interaction on body mass index (BMI) PGS (PGSBMI) 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 R2 differences among strata. Covariates with the largest differences included age, sex, blood lipids, physical activity, and alcohol consumption, with R2 being nearly double between best and worst performing quintiles for certain covariates. 28 covariates had significant PGSBMI-covariate interaction effects, modifying PGSBMI effects by nearly 20% per standard deviation change. We observed overlap between covariates that had significant R2 differences among strata and interaction effects - across all covariates, their main effects on BMI were correlated with their maximum R2 differences and interaction effects (0.56 and 0.58, respectively), suggesting high-PGSBMI individuals have highest R2 and increase in PGS effect. Using quantile regression, we show the effect of PGSBMI increases as BMI itself increases, and that these differences in effects are directly related to differences in R2 when stratifying by different covariates. Given significant and replicable evidence for context-specific PGSBMI performance and effects, we investigated ways to increase model performance taking into account non-linear effects. Machine learning models (neural networks) increased relative model R2 (mean 23%) across datasets. Finally, creating PGSBMI directly from GxAge GWAS effects increased relative R2 by 7.8%. These results demonstrate that certain covariates, especially those most associated with BMI, significantly affect both PGSBMI performance and effects across diverse cohorts and ancestries, and we provide avenues to improve model performance that consider these effects.

Selection, optimization and validation of ten chronic disease polygenic risk scores for clinical implementation in diverse US populations.

Polygenic risk scores (PRSs) have improved in predictive performance, but several challenges remain to be addressed before PRSs can be implemented in the clinic, including reduced predictive performance of PRSs in diverse populations, and the interpretation and communication of genetic results to both providers and patients. To address these challenges, the National Human Genome Research Institute-funded Electronic Medical Records and Genomics (eMERGE) Network has developed a framework and pipeline for return of a PRS-based genome-informed risk assessment to 25,000 diverse adults and children as part of a clinical study. From an initial list of 23 conditions, ten were selected for implementation based on PRS performance, medical actionability and potential clinical utility, including cardiometabolic diseases and cancer. Standardized metrics were considered in the selection process, with additional consideration given to strength of evidence in African and Hispanic populations. We then developed a pipeline for clinical PRS implementation (score transfer to a clinical laboratory, validation and verification of score performance), and used genetic ancestry to calibrate PRS mean and variance, utilizing genetically diverse data from 13,475 participants of the All of Us Research Program cohort to train and test model parameters. Finally, we created a framework for regulatory compliance and developed a PRS clinical report for return to providers and for inclusion in an additional genome-informed risk assessment. The initial experience from eMERGE can inform the approach needed to implement PRS-based testing in diverse clinical settings.

Genetic and Survey Data Improves Performance of Machine Learning Model for Long COVID.

Over 200 million SARS-CoV-2 patients have or will develop persistent symptoms (long COVID). Given this pressing research priority, the National COVID Cohort Collaborative (N3C) developed a machine learning model using only electronic health record data to identify potential patients with long COVID. We hypothesized that additional data from health surveys, mobile devices, and genotypes could improve prediction ability. In a cohort of SARS-CoV-2 infected individuals (n=17,755) in the All of Us program, we applied and expanded upon the N3C long COVID prediction model, testing machine learning infrastructures, assessing model performance, and identifying factors that contributed most to the prediction models. For the survey/mobile device information and genetic data, extreme gradient boosting and a convolutional neural network delivered the best performance for predicting long COVID, respectively. Combined survey, genetic, and mobile data increased specificity and the Area Under Curve the Receiver Operating Characteristic score versus the original N3C model.