Using Machine Learning to Integrate Socio-Behavioral Factors in Predicting Cardiovascular-Related Mortality Risk

Cardiovascular disease is prevalent and associated with significant mortality rate. Robust lifetime risk stratification for cardiovascular disease is important for effective prevention, early diagnoses, targeted intervention, and improved prognosis. Health disparities, manifested as socio-behavioral factors, are believed to have multiple effects throughout life with great complexity. Multiple studies investigated lifetime cardiovascular-related mortality risk prediction focusing on subjects' pathophysiology and intervention profiles. In this study, we applied machine learning algorithms and focused on integrating socio-behavioral factors to pathophysiology and intervention profiles to predict cardiovascular-related mortality risk. Our results showed that multiple machine learning algorithms can predict risk with reasonable accuracy, using mixed types of features. Particularly, socio-behavioral factors contributed significantly to the improved accuracy of mortality risk prediction. Feature analysis identified the odds ratio of socio-behavioral factors for cardiovascular-related mortality and offered potential insights on how they impact subjects' long-term outcomes. Our results call for further investigation of this important topic.