Feedback with clinical text representation.

Health records contain rich sources of mental health data that can be used to evaluate disability and health care outcomes. However, a lack of behavioral health ontologies focused on daily life activity functioning has impeded development of clinical informatic tools to extract mental functioning information. We aim to present the theoretical foundation and conceptual model upon which the Ecological Mental Functioning Ontology (EMFO) was built to facilitate natural language processing (NLP) to extract mental functioning information in free-text clinical records. Subject matter experts operationally defined mental functioning, and a related theoretical perspective was established. Face validity of a proposed model was obtained using an iterative grounded theory approach. An annotation schema based on the model was constructed and tested using manual annotation and consensus on datasets of real and synthetic clinical notes. An annotation schema, based on the Ecological Model of Mental Functioning (EMMF), was shown to be robust when using NLP methods to identify and extract mental functioning information in real and synthetic behavioral health clinical notes. Mental functioning is a complex phenomenon that is fully conceptualized within an ecological milieu encompassing the dynamic transactive relationship between the person, the nature and demands of activities the person participates in, and the external contextual and environmental factors within which the activities take place. By operationalizing mental functioning, the EMMF provided a conceptual roadmap to develop the EMFO and NLP methods that identify and extract mental functioning activity information in clinical records.