The purpose of this award is to provide Dr. Elsie Ross, Assistant Professor of Surgery (Vascular Surgery) and Medicine (Biomedical Informatics Research) at Stanford University, the support necessary to transition her from a junior investigator into an independent surgeon-scientist in translational biomedical informatics. Dr. Ross is a vascular surgeon with an advanced degree in health services research and postdoctoral training in biomedical informatics. Her long-term goal is to combine her interdisciplinary training to develop and implement machine learning tools that will enable the delivery of precise, high-value care to patients with cardiovascular diseases. Her career development activities focus on advancing her ability to translate informatics discoveries into viable clinical tools by 1) completing didactic courses to deepen and expand her knowledge of deep learning algorithms, clinical trials and implementation science, 2) designing and conducting her first independent human subjects clinical research study evaluating the performance of machine learning technology, 3) implementing and evaluating the effects of an electronic health record (EHR)-based screening tool to identify latent vascular disease, and 4) strengthening her previous training in cost-effectiveness analysis to enable her future aim of evaluating the associated costs and utility of pro-active, automated disease screening. The candidate has convened a mentorship team that includes Dr. Nigam Shah, a biomedical informatics expert who combines machine learning, text-mining and medical ontologies to enable a learning health care system; Dr. Kenneth Mahaffey a world-expert in cardiovascular clinical trials; and Dr. Paul Heidenreich, an expert in implementation sciences with a focus on the use of EHR interventions to improve care quality for cardiovascular patients and evaluating the cost-effectiveness of new technologies. The research proposal builds on the candidate's prior work with using machine learning and EHR data to evaluate and predict cardiovascular disease outcomes. The candidate now proposes to characterize the performance of machine learning algorithms in identifying patients with peripheral artery disease (PAD) using EHR data (Aim 1), evaluate whether learned classification models perform better than traditional risk factors for identification of undiagnosed PAD in a prospective patient cohort (Aim 2), and implement an EHR-based screening tool to identify patients with undiagnosed PAD and evaluate the diagnosis and treatment effects (Aim 3). Completion of the proposed research will result in a novel, EHR-based screening tool for identification of undiagnosed vascular disease that can decrease PAD-related cardiovascular morbidity and mortality through earlier and more aggressive medical management. This research will also form the basis for an R01 application before the end of the award to conduct a multi-site randomized-controlled clinical trial to evaluate the impact of EHR- based proactive PAD screening. ! ! !
Peripheral artery disease is a prevalent yet under-diagnosed condition that can lead to limb loss, stroke, heart attacks and/or premature death. Work proposed in this grant aims to develop technology using electronic health records and machine learning algorithms to automatically identify patients with undiagnosed peripheral artery disease and recommend treatment. Such technology could improve the health and longevity of patients with peripheral artery disease by ensuring that patients are diagnosed early and appropriately treated.