Research & Development
Ongoing CRIS projects: suicide
Feasibility, acceptability and validation of OxMIS, a novel suicide risk assessment tool: using natural language processing in CRIS for external validation
Oxford Mental Illness and Suicide tool (OxMIS) is a digital suicide risk assessment model for patients with severe mental illness (schizophrenia-spectrum disorders or bipolar disorder). It was initially developed in Oxford using a research dataset from Sweden. However, before using it on real medical records data in the UK, we need to test the model on a dataset collected in the UK to prove it can be generalised and perform to the same high standard as on the Swedish data. This is called external validation. Electronic health records in UK CRIS provide such a possibility. However, some variables used by OxMIS are not structuralised in the UK CRIS dataset. Instead, they are hidden within free-text that cannot be directly used or fed into the model. Thus, we need a tool named Natural Language Processing (NLP) to extract missing structured information from clinical notes. In this study, we examined the feasibility of identifying such information needed with the NLP tool to validate the OxMIS model in routinely collected electronic health records.