• A POSTnote on AI and healthcare will look at the current applications, opportunities and limitations of AI.
  • It will review available evidence on the impact of AI on healthcare delivery and costs and will explore the main ethical, social and regulatory challenges to its use.
  • Provisional start date: September 2020. To contribute expertise, literature or an external reviewer please contact Dr Lorna Christie. View our guidance for expert contributors.

Artificial intelligence (AI) can be used in healthcare settings in a number of ways, including analysis of medical images and other data. It can assist in diagnoses, scanning scientific literature and patient data to help inform decisions about treatment. It can provide ‘virtual’ healthcare assistants and use medical data to make predictions about disease progression.

It has also been suggested that AI could be used to automate NHS administration processes and schedule resources more efficiently. The use of AI in the NHS is not widespread. However, trials of the technology have taken place across the UK.

For example, a trial carried out by Moorfields Eye Hospital and the company DeepMind showed that an AI system can use eye scan data to detect around 50 eye conditions and make a correct referral decision with 94% accuracy (matching the accuracy of clinical experts).

Another AI trial at Hammersmith Hospital predicted the prognosis of patients with ovarian cancer more accurately than current methods.

The use of AI in healthcare raises several issues, including the potential for AI to make erroneous decisions, and challenges concerning where clinical accountability lies in processes that incorporate AI. Widespread implementation of AI in the NHS would also require access to large amounts of medical data, raising issues around data privacy and consent.

This POSTnote will provide an overview of AI in healthcare including its current applications, opportunities and limitations. It will review available evidence on the impact of AI on healthcare delivery and costs and explore the main ethical, social and regulatory challenges to its use.