Maame Yaa “Maya” A. B. Yiadom

We are developing a predictive model suited for emergency clinical care, that incorporates risk differences associated with race, to achieve algorithmic equity over structured racism within current diagnostic practice. Our work will help us better understand this observed racial disparity in timely diagnosis that is resulting in an emergency care access barrier disproportionately affecting patients of non-white race. Our goal is to improve the sensitivity of diagnosing all STEMI patients while reducing long standing disparities across racial groups. We do this by pursuing algorithmic equity by using ACS predicting artificial intelligence (AI) to supplement human driven practice.

We hypothesize a, race-, sex-, ethnicity-, and language-inclusive AI screening model will identify patients in need of an early ECG with greater sensitivity and precision than current practice.

International guidelines recommend an early ECG for patients with symptoms suggestive of ACS within 10 minutes of emergency department (ED) arrival to identify the subset with STEMI. Symptoms for STEMI overlap with those of other types of ACS. Patients with STEMI who do not receive an ECG within 10 minutes have twice the 1-week mortality of those who do: 10.2% vs 5%.

Patients of non-white races are over-represented among those who have a late diagnosis of ST-segment elevation myocardial infarction (STEMI), the most severe form of acute coronary syndrome (ACS). Our prior work identified that STEMI patients whose early ECG were >10 minutes were disproportionately of a non-white race (87% vs. 65%, p=0.028) and treated with PCI later (207 vs. 93 minutes, p=0.048) than those receiving a timely ECG. The completion of this work will help us to optimize the selection of all patients upon ED arrival who need an ECG for more timely treatment.