Data integration and interpretability of electronic health records with OMOP

Abstract

Medical centers such as clinics, hospitals, health centers, and service providers generate and consume a large amount of data in their daily activities. This data is stored in Electronic Health Records (EHR) that store a wide spectrum of patient data including medical history, laboratory test results, drugs administered, and diagnostics. This information is of great importance to support the physician's decisions when evaluating a patient, providing diagnoses, and proposing medical treatments. With the use of data mining methods, it is possible to extract useful knowledge and support the physician's decision, improving the quality of their diagnoses and the treatments prescribed. The objective of this research project is to develop data mining algorithms to cluster, detect, and explain patterns and anomalies in EHR databases, especially those integrated with OMOP. Thus, the methods to be proposed are aimed at helping physicians providing accurate diagnoses and indicating effective treatments by exploring heterogeneous medical databases. Also, the new approuches can be contribute to development of assistive and rehabilitation technologies to people with severe disabilites that allow increasing their quality of life. (AU)