Using Predictive Modeling for Acute Coronary Syndrome Screening to Improve Timely Diagnosis and Mortality for STEMI - PROJECT SUMMARY / ABSTRACT We propose that artificial intelligence (AI) screening for acute coronary syndrome (ACS) can improve ST-elevation myocardial infarction (STEMI) identification, and eliminate structured biases that disproportionately cause diagnostic delay in women and non-white patients. Every year 26 million patients visit EDs for ACS evaluation. All need an ECG in a critical 10-minute window of time after arrival to diagnose the most time-sensitive and life-threatening form of ACS, STEMI. These early ECGs occur before physician evaluation. In our prior work, every 1-minute reduction in time to ECG yielded a 1.24-minute reduction in the time to STEMI treatment, and every minute improves ACS clinical outcomes. Yet 37% of STEMI patients do not receive an ECG within 10-minutes. Those experiencing ECG delay disproportionately include women and non-white patients. These patients are twice as likely to die within 1 week compared to those receiving an ECG within the 10-minute window (11% vs 5%, p=0.02). These unacceptable findings need a diagnostic improvement solution. We will address this challenge by developing diversity-inclusive ACS screening artificial intelligence (AI) that will support ED care to 1) prevent missing STEMI patients by improving on the existing oversimplified manual screening practice, and 2) eliminate structured biases disproportionately impacting women and non-white patients for whom we will improve diagnostic precision. Our proposal aims to reduce disparities in timely care and mortality to improve equity via 2 innovative contributions. First, we will test the desktop-to-bedside translation of a predictive model that out-performs manual screening. Second, we will improve the model by including considerations for risk diversity and leveraging the calculation strength of machine learning to improve ACS prediction. Our goal is to reduce mortality for all by bridging a translational research gap. The impact of overcoming these practice limitations is that we will validate a pathway to reduce delays in STEMI identification. We will also close a clinical outcome disparity between women and men, and patients of non-white and white races. Our multidisciplinary team of experts in emergency medicine, interventional cardiology, predictive modeling, biostatistics, and clinical informatics have the expertise to successfully execute this proposal.
