Fig. 1

Natural language processing (NLP) case examples using MedCAT (medical concept annotation tool). (a) Surfacing vital clinical patient information to enable research and drive clinical insights (b) Visualising and forecasting patient timelines (c) personal health information redaction of electronic health records. (a) Using MedCAT to surface distribution of physical (Left) and mental disorders (Right) from MIMIC-III (an intensive care unit dataset) [8]. This approach was also used during the COVID-19 pandemic to elucidate risk factors and relationships with medications (ACE inhibitors) to help address international research questions [2]. (b) Visualising patient timelines using MedCAT enables us to better understand disease trajectories as well as public health planning. Patient timeline data can then be used to train an AI model (generative pretrained transformer) to predict the next probable clinical event. (Image from publication Zeljko et al.) [6] Creative Commons Attribution (CC BY 4.0). (c) Using MedCAT with clinician fine-tuning, we are able to redact personal health information to preserve patient privacy. We have de-identified and stored over 2 million free-text documents in King’s College Hospital using this approach, this enables safer data-sharing approaches for future research and operational projects [5]