Using machine learning to understand cardiopulmonary and renal end organ complications for individuals with SCD

PROJECT SUMMARY/ABSTRACT Individuals with sickle cell disease (SCD) are at risk of developing numerous chronic complications including cardiopulmonary and renal (CPR) end organ complications, which are associated with early mortality. To better understand the risks of CPR end organ complications and outcomes of individuals with SCD and CPR end organ complications, large epidemiological studies are needed. However, given that SCD is a rare condition, there remains a lack of large epidemiological studies focused on CPR end organ damage among individuals with SCD. In this project, we propose to establish a large patient cohort leveraging existing big data sources of the electronic health record (EHR) and administrative claims to determine the risks of CPR end organ complications and factors associated with frequent acute care use among those with CPR end organ complications and SCD. In this project, the candidate will obtain training in machine learning models used for risk prediction along with training for the causal inference framework and recently developed counterfactual based machine learning ensembles. The candidate has established a mentoring team with content expertise in SCD, health care services research, and machine learning mentor. In addition, the scientific advisory committee includes members that cross discipline and complement the mentoring team. The skill sets acquired during the award period will be directly applied to the proposed project and help address many of the unanswered questions in the field of SCD.