Collaborative Prediction Model of Disease Risk by Mining Electronic Health Records

Patient Electronic Health Records (EHR) is one of the major carriers for conducting preventative medicine research. However, the heterogeneous and longitudinal properties make EHRs analysis an inherently challenge. To address this issue, this paper proposes CAPM, a Collaborative Assessment Prediction Model based on patient temporal graph representation, which relies only on a patient EHRs using ICD-10 codes to predict future disease risks. Firstly, we develop a temporal graph for each patient EHRs. Secondly, CAPM uses hybrid collaborative filtering approach to predict each patient’s greatest disease risks based on their own medical history and that of similar patients. Moreover, we also calculate the onset risk with the corresponding diseases in order to take action at the earliest signs. Finally, we present experimental results on a real world EHR dataset, demonstrating that CAPM performs well at capturing future disease and its onset risks.