Dataset Information

Quantitative disease risk scores from EHR with applications to clinical risk stratification and genetic studies.

ABSTRACT: Labeling clinical data from electronic health records (EHR) in health systems requires extensive knowledge of human expert, and painstaking review by clinicians. Furthermore, existing phenotyping algorithms are not uniformly applied across large datasets and can suffer from inconsistencies in case definitions across different algorithms. We describe here quantitative disease risk scores based on almost unsupervised methods that require minimal input from clinicians, can be applied to large datasets, and alleviate some of the main weaknesses of existing phenotyping algorithms. We show applications to phenotypic data on approximately 100,000 individuals in eMERGE, and focus on several complex diseases, including Chronic Kidney Disease, Coronary Artery Disease, Type 2 Diabetes, Heart Failure, and a few others. We demonstrate that relative to existing approaches, the proposed methods have higher prediction accuracy, can better identify phenotypic features relevant to the disease under consideration, can perform better at clinical risk stratification, and can identify undiagnosed cases based on phenotypic features available in the EHR. Using genetic data from the eMERGE-seq panel that includes sequencing data for 109 genes on 21,363 individuals from multiple ethnicities, we also show how the new quantitative disease risk scores help improve the power of genetic association studies relative to the standard use of disease phenotypes. The results demonstrate the effectiveness of quantitative disease risk scores derived from rich phenotypic EHR databases to provide a more meaningful characterization of clinical risk for diseases of interest beyond the prevalent binary (case-control) classification.

SUBMITTER: Xu D

PROVIDER: S-EPMC8302667 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:OBJECTIVES:This study assessed the utility of the pooled cohort equation (PCE) and/or coronary artery calcium (CAC) for atherosclerotic cardiovascular disease (ASCVD) risk assessment in smokers, especially those who were lung cancer screening eligible (LCSE). BACKGROUND:The U.S. Preventive Services Task Force recommended and the Centers for Medicare & Medicaid Services currently pays for annual screening for lung cancer with low-dose computed tomography scans in a specified group of cigarette smokers. CAC can be obtained from these low-dose scans. The incremental utility of CAC for ASCVD risk stratification remains unclear in this high-risk group. METHODS:Of 6,814 MESA (Multi-Ethnic Study of Atherosclerosis) participants, 3,356 (49.2% of total cohort) were smokers (2,476 former and 880 current), and 14.3% were LCSE. Kaplan-Meier, Cox proportional hazards, area under the curve, and net reclassification improvement (NRI) analyses were used to assess the association between PCE and/or CAC and incident ASCVD. Incident ASCVD was defined as coronary death, nonfatal myocardial infarction, or fatal or nonfatal stroke. RESULTS:Smokers had a mean age of 62.1 years, 43.5% were female, and all had a mean of 23.0 pack-years of smoking. The LCSE sample had a mean age of 65.3 years, 39.1% were female, and all had a mean of 56.7 pack-years of smoking. After a mean of 11.1 years of follow-up 13.4% of all smokers and 20.8% of LCSE smokers had ASCVD events; 6.7% of all smokers and 14.2% of LCSE smokers with CAC = 0 had an ASCVD event during the follow-up. One SD increase in the PCE 10-year risk was associated with a 68% increase risk for ASCVD events in all smokers (hazard ratio: 1.68; 95% confidence interval: 1.57 to 1.80) and a 22% increase in risk for ASCVD events in the LCSE smokers (hazard ratio: 1.22; 95% confidence interval: 1.00 to 1.47). CAC was associated with increased ASCVD risk in all smokers and in LCSE smokers in all the Cox models. The C-statistic of the PCE for ASCVD was higher in all smokers compared with LCSE smokers (0.693 vs. 0.545). CAC significantly improved the C-statistics of the PCE in all smokers but not in LCSE smokers. The event and nonevent net reclassification improvements for all smokers and LCSE smokers were 0.018 and -0.126 versus 0.16 and -0.196, respectively. CONCLUSIONS:In this well-characterized, multiethnic U.S. cohort, CAC was predictive of ASCVD in all smokers and in LCSE smokers but modestly improved discrimination over and beyond the PCE. However, 6.7% of all smokers and 14.2% of LCSE smokers with CAC = 0 had an ASCVD event during follow-up.

Dataset Information

Quantitative disease risk scores from EHR with applications to clinical risk stratification and genetic studies.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets