Question to the Department of Health and Social Care:

To ask His Majesty's Government what assessment they have made of the use of predictive AI models in NHS England to forecast when someone is likely to become ill.

Against a backdrop of increasing demand for National Health Services and significant workforce pressures, artificial intelligence (AI) presents significant opportunities for improving the delivery of care and outcomes for patients, whilst also saving money and staff time. AI even has the potential to support with the prediction and prevention of illness.

AI tools that can assist with predicting the health trajectory of patients have been trialled in multiple NHS England trusts. These tools use existing healthcare data to recognise complex patterns and forecast future disorders, symptoms, medications, and procedures. These tools have been trained on de-identified, anonymised NHS electronic health records, which ensures that patient data is protected whilst allowing the AI to learn from real United Kingdom health data that represents our diverse population. The initial trials of these predictive AI tools showed accuracy levels of 68% and 76% in two major NHS England trusts when forecasting the next possible disorders that could appear in a patient’s timeline. These accuracy levels rose to 80% and 81% when predicting wider concepts, including disorders, symptoms, relapses, or medication. These trials were funded by the NHS AI Lab and the National Institute for Health and Care Research, and were closely monitored by the Department.

Deployment of AI in the NHS is still at a relatively early stage, with the majority of AI technologies, including predictive models, being deployed in a research capacity. To address this, the Department is carrying out work to assess the barriers of effective adoption and improve the way AI tools are deployed across the NHS.