Back

An explainable AI approach for discovering social determinants of health and risk interactions for stroke in patients with atrial fibrillation

Zimmerman, R. M.; Hernandez, E. J.; Watkins, W. S.; Blue, N.; Tristani-Firouzi, M.; Yandell, M.; Steinberg, B. A.

2023-03-01 cardiovascular medicine
10.1101/2023.02.27.23286470 medRxiv
Show abstract

Atrial fibrillation (AF) leads to significant morbidity and mortality, which is primarily related to stroke despite effective stroke prevention therapies. There remains a critical need for personalized, socially aware, equitable stroke risk prediction among patients with AF to enable optimal implementation of contemporary stroke-prevention therapies. In this brief report, we leverage innovative computational tools and high-quality, extensive data (1.8 m patients, augmented with social determinants of health information) to demonstrate the ability of a unique, explainable AI approach to improve the accuracy and equity of stroke risk prediction. Current risk stratification approaches are blind to social determinants of health and fail to adjust for unique contributions and interactions of variables upon stroke risk. In contrast, our results indicate that social determinants of health can be important modifiers of clinical variables and ultimately stroke risk. We hope that this analysis can provide evidence to drive better, more personalized, and equitable stroke risk stratification and prevention for patients with AF in the future.

Matching journals

The top 10 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.