Precision Cardio-Metabolic Phenotyping for Genetic Discovery and Risk Prediction

2型糖尿病（T2D）和心血管疾病（CVD）是发病的主要原因， 美国退伍军人的死亡率，以及整个美国人口。2型糖尿病是一个广泛认可的心血管疾病的危险因素， 而T2D会导致更糟糕的CVD结果。然而，在这些患者中仍存在相当大的临床异质性。 T2D患者。即使在血糖控制明显相似的个体中， 对于谁将发展CVD的可变性。为了制定更有效的策略来预防心血管疾病， 对于高危人群，需要更好的方法来量化CVD风险。使用新的计算 方法，我们将考虑密集的表型和基因型数据，以确定个体的亚群 2型糖尿病患者是心脏和血管疾病的高危人群。目标1：传统 心血管疾病的危险因素，如胆固醇、血压和吸烟，以及三种心脏和血管疾病 表型：外周动脉疾病（PAD）、冠心病（CHD）和脑血管疾病， 得到考验为了解释这些结果经常发生在同一个人身上的事实，统计模型 将特征视为相关的人内结果将被使用。为了确定基因的添加是否 信息改善了CVD结果的预测，遗传风险评分的影响，基于初步的 研究从VA百万退伍军人计划和其他出版的工作，对模型将被评估。在目标2中 密集的表型数据将从电子健康记录中提取， 双聚类算法将用于识别T2D的隐藏亚型。这些亚型与 然后将评估CVD结果。在目标3中，将采用类似的方法来阐述T2D亚型 基于已知与T2D，CVD及其风险因素相关的DNA变异。最后，基因和 将共同考虑表型数据。这些方法将应用于美国退伍军人的数据， 使用退伍军人老龄化队列研究和VA人群（通过企业数据仓库），以及 非退伍军人，使用来自PennMedicine BioBank，Penn Data Store和UK Biobank的数据。成功 该项目的完成将有助于阐明T2D的表型结构，并确定在 T2D风险最高这些结果将为制定量身定制的CVD预防策略奠定基础 并帮助实现心脏和血管疾病精准医疗的承诺。

Type 2 diabetes (T2D) and cardiovascular disease (CVD) are among the leading causes of morbidity and mortality in US Veterans, as well as the US population at large. T2D is a widely-recognized risk factor for CVD, and T2D leads to worse CVD outcomes. However, there remains considerable clinical heterogeneity among individuals with T2D. Even among individuals with apparently similar glycemic control, there is significant variability with respect to who will develop CVD. To develop more effective strategies to prevent CVD in this high-risk population, better approaches for quantifying CVD risk are needed. Using novel computational approaches, we will consider dense phenotype and genotype data to identify the subpopulations of individuals with T2D who are at the highest risk of heart and vascular disease. In Aim 1, the relationship between traditional CVD risk factors, such as cholesterol, blood pressure, and smoking, and three heart and vascular disease phenotypes: peripheral artery disease (PAD), coronary heart disease (CHD), and cerebrovascular disease, will be tested. To account for the fact that these outcomes frequently occur in the same individuals, statistical models that treat the traits as correlated-within person outcomes will be used. To determine if the addition of genetic information improves the prediction of CVD outcomes, the impact of genetic risk scores, based on preliminary studies from the VA Million Veteran Program and other published work, on the models will be assed. In Aim 2 dense phenotype data will be extracted from the electronic health record and novel artificial intelligence based biclustering algorithms will be used to identify hidden subtypes of T2D. The association of these subtypes with CVD outcomes will then be assessed. In Aim 3, a similar approach will be taken to elaborate T2D subtypes based on DNA variants known to associate with T2D, CVD, and their risk factors. Finally, the genetic and phenotypic data will be jointly considered. These approaches will be applied across data from both US Veterans, using the Veterans Aging Cohort Study and the VA population at large (via the Corporate Data Warehouse), and non-Veterans, using data from the PennMedicine BioBank, Penn Data Store, and UK Biobank. Successful completion of this project will help to elucidate the phenotype structure of T2D and identify individuals at the highest risk of T2D. These results will lay the ground work for developing tailored strategizes for CVD prevention in T2D and help realize the promise of precision medicine for heart and vascular disease.