NHS England has paused a ground-breaking AI project designed to predict an individual’s risk of health conditions after concerns were raised data from 57 million people was being used without the right permissions.
Foresight, which uses Meta‘s open-source AI model, Llama 2, was being tested by researchers at University College London and King’s College London as part of a national pilot scheme exploring how AI could be used to tailor healthcare plans for patients based on their medical history.
But the brakes were applied to the pioneering scheme after experts warned even anonymised records could contain enough information to identify individuals, The Observer reported.
A joint IT committee between the British Medical Association (BMA) and the Royal College of General Practitioners (RCGP) also said it they had not been made aware that data collected for research into Covid was now being used to train the AI model.
The bodies have also accused the research consortium, led by Health Data Research UK, of failing to consult an advisory body of doctors before feeding the health data of tens of millions of patients into Foresight.
Both BMA and RGCP have asked NHS England to refer itself to the Information Commissioner over the matter.
Professor Kamila Hawthorne, chair of RGCP, said the issue was one of ‘fostering patient trust’ that their data was not being used ‘beyond what they’ve given permission for.’
She said: ‘As data controllers, GPs take the management of their patients’ medical data very seriously, and we want to be sure data isn’t being used beyond its scope, in this case to train an AI programme.
Hellbound and Down 