Posted on April 4, 2018 by staff

NHS investing in AI avatar to improve data access


The NHS is investing in an artificial intelligence avatar which will allow people to access their healthcare data more easily.

NHS Digital’s director of data Professor Daniel Ray told BusinessCloud that the IPsoft avatar – known as Amelia – could be the first point of contact for patients seeking personal information.

Available 24/7, it would also enable clinicians, researchers and commissioners to find and appropriately access the data they need and therefore improve NHS services, according to Prof Ray.

“AI can absolutely play a role in making healthcare more cost-effective and affordable, given the burden the population is placing on the NHS,” he said at the HIMSS UK Executive Leadership Summit at the King’s Fund in London.

“At the moment we’ve got people who are helping navigation to the data that we hold, but we’re investing in technology which will allow an avatar to do that. If you want some A&E data [for example] you’ll come and talk to ‘Amelia’.

“You can still talk to a person, of course, but we’re going to put technology up front.”

Requests through NHS Digital for all types of data are increasing, with double the number of applications received in 2017 compared with 2016. A little more than half of applications are for NHS purposes, with a quarter for research or academic purposes and a quarter for other national health bodies.

An avatar could make the process faster and more efficient while not losing the checks and balances necessary to keep information secure.

“The implementation of AI will be in a staged approach. We are at the early stages of it at the moment,” Prof Ray continued.

“There are still a number of unanswered questions such as regulation, commercial IP, ownership, ongoing tracking, and who does the AI make its decisions in favour of – is it the health system, or always the patient?

“It needs to walk before it can run, focusing in on making some of the routine decisions that are taken in a reliable way.

“Once it’s proven that it can do that efficiently and effectively without any adverse consequences and we maintain patient trust, we can start to advance it into slightly more complex areas.”

He added: “We’ve got a challenge around communicating the benefit to patients of AI. We need to come up with a set of language and phrases around some key points that we can put out as to what the true and real benefits are.

“Let’s not try and explain what’s in the black box, but try to put what it does in layman’s terms: how it may get key decisions made more reliably, free doctors up to do more complex things, take away some of the mundane routine things and improve the accuracy of diagnostics.

“These are easy for people to understand.”