Application of concise machine learning to construct accurate and interpretable EHR computable phenotypes

Abstract

ObjectiveElectronic health records (EHRs) can improve patient care by enabling systematic identification of patients for targeted decision support. But, this requires scalable learning of computable phenotypes. To this end, we developed the feature engineering automation tool (FEAT) and assessed it in targeting screening for the under-diagnosed, under-treated disease primary aldosteronism.Materials and MethodsWe selected 1199 subjects receiving longitudinal care in one health system between 2007 and 2017 and classified them for hypertension (N=608), hypertension with unexplained hypokalemia (N=172), and apparent treatment-resistant hypertension (N=176) by chart review. We derived 331 features from EHR encounters, diagnoses, laboratories, medications, vitals, and notes. We modified FEAT to encourage model parsimony and compared its models’ performance and interpretability to that of expert-curated heuristics and conventional machine learning.ResultsFEAT models trained to replicate expert-curated heuristics had higher AUPRC scores than all other models (p < 0.001) except random forests and were smaller than all other models (p < 1e-6) except decision trees. FEAT models trained to predict chart review phenotypes exhibited similar AUPRC scores to penalized logistic regression while being substantially simpler than all other models (p < 1e-6). For treatment-resistant hypertension, FEAT learned a six-feature, clinically intuitive model that demonstrated an adjusted PPV of 0.73 and sensitivity of 0.54 in testing.DiscussionFEAT learns computable phenotypes that approach the performance of expert-curated heuristics and conventional machine learning without sacrificing interpretability.ConclusionBy constructing accurate and interpretable computable phenotypes at scale, FEAT has the potential to facilitate widespread, systematic clinical decision support.