NHS team develops AI tool for health predictions

A group of healthcare researchers within the NHS has created an AI tool called Foresight to forecast the health journey of patients. Foresight has the potential to aid in clinical decision-making, healthcare monitoring, and clinical trials.

The team, comprising researchers from King’s College Hospital, Guy’s and St Thomas’ NHS foundation trusts, King’s College, and University College London, developed Foresight using the Cogstack platform. This platform utilises natural language processing to analyse NHS electronic health record data. Foresight employs deep learning techniques to recognize intricate patterns in both structured and unstructured data.

According to a study published in The Lancet Digital Health, Foresight demonstrated promising results, correctly predicting the next 10 potential disorders in patient timelines with a success rate of 68% at King’s College Hospital, 76% at Maudsley NHS Foundation Trust, and 88% in the US MIMIC-III dataset.

In a simulation assessment, five clinicians created 34 mock patient timelines with various scenarios, and Foresight accurately predicted relevant outcomes 93% of the time, as judged from a clinical perspective.

A report from the NIHR Maudsley Biomedical Research Centre suggested that Foresight could be utilised for real-world risk forecasting, emulating clinical trials, and studying disorder progression.

Professor Richard Dobson, senior author of the study and professor of medical informatics at King’s College London, highlighted the wide-ranging applications of Foresight, including digital health twins, synthetic dataset generation, longitudinal research, and medical education.

Funding for the study was provided by the NHS AI Lab, the National Institute for Health and Care Research, and Health Data Research UK (HDRUK).

Professor Andrew Morris, director of HDRUK, emphasised the importance of investing in the UK's data infrastructure to ensure the safe and secure utilisation of electronic health records and AI advancements for enhancing clinical decision-making.

Click here to read the original news story.