From observation to evidence: harnessing the mental status examination as a real-world data source in clinical research.

Background: Electronic health record (EHR) based research using psychiatric real-world data is limited by a lack of measurement-based care. The mental status examination (MSE) offers a routinely documented, alternative data source for clinical phenotyping, potentially addressing this evidence gap. Aim: This article aims to provide a comprehensive clinical and scientific justification for the use of MSE data as a source of real-world evidence, and to critically examine the assumptions underlying its use. Discussion: The content and structure of the MSE are first described and compared with those of structured psychometric rating instruments. Current literature applying natural language processing (NLP) methods, (including large language models (LLMs)), to extract psychopathological information from the EHR is briefly reviewed. Schizophrenia is then used as a case study for an in-depth critical examination of the assumptions and complexities inherent in mapping MSE findings to their corresponding psychopathological constructs. Principles for subject-matter-expert review of MSE data are proposed alongside a sample conceptual organization of MSE findings according to the main psychopathological domains of schizophrenia. Contributions made by MSE data at the research frontier are discussed, and the limitations of MSE data are considered.