Project summary Chronic kidney disease (CKD) is common, affecting 14.8% of US adults, and disproportionately more in diverse and underserved communities. CKD significantly reduces life expectancy and quality of life, while imposing tremendous economic burden on society. A critical need persists for early identification of modifiable risk factors in susceptible populations and to establish actionable support for medical decision making. Among the modifiable risk factors, drug induced acute kidney injury (AKI) contributes to CKD development and progression. The current knowledge of nephrotoxic drug-drug interactions (DDIs) is insufficient to prevent harm in heterogenous patient subpopulations. Electronic health records (EHRs) from electronic medical records (EMR) and health insurance claims data can help predict disparate CKD progression trajectories and uncover novel nephrotoxic drug interactions. The Indiana University School of Medicine (IUSM) EHR collection includes rich clinical information for 38 million individuals from regional and national populations over two-to-three decades. The IUSM EHR collection is composed of Optum EHR derived from the Optum Clinformatics™ claim data and the Indiana EHR incorporated from the EMR data of Indiana Network for Patient Care (INPC) Research Database, Indiana University Health (IUH), and Eskenazi Health (EH). We propose to develop the DisEase PrOgression Trajectory (DEPOT), an evidence-driven, graph-based clinical informatics approach to model CKD progression trajectories and individualize clinical decision support. We hypothesize that there are different CKD progression paths which are: 1) driven by different pathogenic mechanisms, 2) susceptible to different nephrotoxic drugs, and 3) identified by unique EHR data patterns. Mathematically, such CKD trajectory landscapes can be learned as principle graphs representing the topological and temporal characteristics of the observed, fragmented EHR data. The goal of this work is to use the IUSM EHR data collection to 1) establish EHR-based CKD progression trajectories and 2) to learn actionable knowledge to prevent drug-induced AKI and CKD. The multi-specialty team proposes to: Aim 1. Construct CKD progression trajectories using graph artificial intelligence model and the IUSM EHR data and Aim 2) Identify nephrotoxic DDIs in the general population and trajectory-specific subpopulations that increase risks of AKI and CKD. The success of the proposed work will generate novel knowledge about the landscape of CKD health trajectories and nephrotoxic DDIs, bridging gaps between rich longitudinal EHR data and decision support for precision medicine in CKD. This work will shift paradigms of big data and complex disease research, enabling EHR data to become part of daily CKD management.