A doctor of philosophy has explained why regulation and infrastructure needs to be in place before artificial intelligence (AI) can be deployed in medicine.

Natalie Banner, who works as Understanding Patient Data Lead & Policy Adviser at the Wellcome Trust, was considering the question of when AI could impact healthcare in the UK at this morning’s AIMed Breakfast meeting A CLINICIAN’S GUIDE TO THE PRACTICAL APPLICATIONS OF AI IN MEDICINE’.

Dr Banner highlighted some of the dangers of implementing AI before the appropriate infrastructure of legislation and regulation was in place to control new innovations with this technology.

She said: “There are fantastic pockets of innovation happening all over the world, in the UK, the US, everywhere.”

Watch the full video of the briefing by signing up to our resources page here.

“On a local level, really exciting developments and innovations are springing up.”

“What you don’t want to do is create a unifying system which controls everything from the top down.

“We all know the disadvantages of trying to control your IT infrastructure from a top down level.

“However, the problem with having no top down control is those little pockets of innovation are just going to remain in the areas where they’re being created.”

“That creates a real risk of perpetuating health inequalities.

“If only some areas adopt the latest innovations, then you develop a postcode (zip code) lottery where some areas can get the benefits and some can’t.”

“Additionally, if you’re only training AI on particular populations then it might not necessarily be representative, you introduce issues of bias, etc.

“Conceivably in the next few years we could get a lot more AI innovations cropping up all over the place but at a small level.

“I think it would be an absolute opportunity missed if the enabling infrastructure isn’t put into place so the best and most useful innovations can actually proliferate and scale and benefit far broader populations.”

Watch the full video of the briefing by signing up to our resources page here.

See more meeting REVIEW reports

MEETING REVIEW: HOW NHS COULD BE EFFICIENT WITH AI FOR MAMMOGRAM

MEETING QS: GETTING PUBLIC ACCEPTANCE OF AI IN THE NHS