Evaluating openEHR for storing computable representations of electronic health record-driven phenotyping algorithms
Electronic Health Records (EHR) are structured and unstructured data generated during routine clinical care. EHR offer researchers unprecedented phenotypic breadth and depth and have the potential to accelerate the pace of translation and enable precision medicine at scale. One of the main use-cases of EHR is creating algorithms to define disease status, onset and severity from diagnoses, prescriptions, laboratory tests, symptoms and other EHR elements - a process known as phenotyping. Currently, no common standardized, structured, computable format exists for defining algorithms and the majority of algorithms tend to be stored as human-readable descriptive text documents. This makes their translation to code ambiguous and challenging and their sharing across the scientific community problematic. In this paper, we evaluate openEHR, a formal open-source EHR data specification, against a list of a priori defined desirable characteristics for storing computable representations of EHR-driven phenotyping algorithm.