Heart failure (HF) is a global epidemic at present and is projected to increase in the future. Despite the critical needs, HF drug discovery efforts are declining because of the requirement of large and long clinical trials to validate efficacy in morbidity and mortality endpoints. A recently issued FDA draft guidance, however, clarified that the effect on symptoms or physical function without a favorable effect on survival or hospitalization could be a basis for approving drugs to treat heart failure. Developing start diagnostic tools to monitor functional properties of cardiovascular systems, including myocardium function, will support a new trend of HF drug development. Annually >1 million diagnostic catheterizations have been performed; numerous data are stored in the electronic health record. Analyzing those data could provide an unprecedented opportunity to identify patterns of functional changes in the hemodynamics of various HFs. Our approach will combine computational and machine learning methods to achieve this goal. Diabetes mellitus (DM) in men and women have a 2X and 4X, respectively, higher risk of heart failure (HF) incident. A recent phenogrouping study identified a subgroup of HFpEF with diabetes having the highest risk of cardiovascular death and hospitalization among other groups. This study brought up an opportunity to develop a targeted therapy for HFpEF with DM (dHFpEF) by developing diagnostic tools to identify candidate dHFpEF patients. Here, we will test the feasibility of diagnostic tools to stratify HFpEF and model it in vitro. Aim 1 of the study is to optimize an already developed diagnostic platform, AI-Assisted, Systems-biology Integrated patient Stratification Technology (AASIST), to analyze data collected by trans-thoracic echocardiography (TTE) and right heart catheterization (RHC) for the purpose of classifying dHFpEF phenotypes. We expect to identify a few groups of dHFpEF defined by their mechanical properties of the myocardium (e.g., elevated left ventricular stiffness). Aim 2 of this study is to analyze corresponding parameters of LV stiffness and contractility in vitro using engineered heart tissues, NuHeart, reconstituted derived from DM patients’ cells. We will culture NuHeart with various environmental challenges to model dHFpEF. While lifestyle risks (e.g., smoking, low physical activities) may outweigh genetic influences, we hypothesize that NuHeart derived from a diabetic patient is susceptible to develop HFpEF phenotype depending on its culture conditions.