Software 'performs better than humans' in Royal College's GP exam

Organisation: Position: Deadline Date: Location:

ArtificialintelligenceIts commercial artificial intelligence diagnostics software performs better than humans in the Royal College of General Practitioner’s exam that must be passed to become a qualified GP doctor in the UK an AI company claims.

Babylon Healthcare Services says that its artificial intelligence software, in tests, can assess common conditions more accurately than human doctors. According to Business Report, the mobile medical consultation service said its AI correctly answered 81% of diagnostic questions designed to mimic those trainee doctors must answer as part of the Royal College of General Practitioner’s exam that must be passed to become a qualified GP doctor in the UK. The exam is graded on a curve, but over the past five years, the average score trainees needed to pass was 72%.

But Martin Marshall, vice chair of the Royal College of General Practitioners, which administers the exam Babylon's AI sought to benchmark itself against, said that software should never be compared to the abilities of human doctors. “No app or algorithm will ever be able to do what a GP does,” Marshall is quoted in the report as saying. “Much of what a GP does is based on a trusting relationship between a patient and a doctor and research has shown that GPs have a 'gut feeling' when they just know something is wrong with a patient.”

It’s not the first time Babylon has tangled with the RCGP either, the report says. The professional body has criticised the UK National Health Service for letting patients choose Babylon's "GP in Hand" mobile consultation service as their official primary care provider. It’s also attacked Babylon for paying doctors who do consultations through its app too much money, making it more difficult for local NHS offices to find doctors to cover shifts.

Babylon demonstrated its AI technology publicly for the first time an event at London’s Royal College of Physicians and said the company would publish the results of its tests online. It’s made the tool available for free through its app and website in some parts of the UK, including London, as well as Rwanda, although because the technology has not been approved by regulators, Babylon calls the software’s answers "health information" not "diagnoses."

The report says in addition to Babylon, others racing to create general diagnostic software include Ada, a start-up with offices in Berlin and London that has launched a similar symptom-checker app to Babylon, HealthTap, in Palo Alto, California, Your.MD, a health information app created by UK medical publisher BMJ, and International Business Machines Corp.

So far, AI companies’ efforts to change health care have not always been smooth. The report says IBM’s Watson Health business has been criticised for falling short of promises that its system could help doctors select the best available cancer treatments for a given patient. Meanwhile, UK regulators faulted a London hospital for transferring millions of patient records to Alphabet Inc’s DeepMind.

Malcolm Grant, chair of NHS England, took part in a panel discussion during the launch event Babylon held to announce the AI test results. He said Babylon's "GP at Hand" model was "interesting," but also sounded a note of caution. "The NHS has enormous intellectual capital and I want to be very clear about not driving that away," Grant said. "I want to be very clear that we will not drive GPs out of practice and see this as an Uber-style rival."

The report says there are also concerns about exactly how tools like Babylon’s will be used. "There is already pushback from patients that they are not getting enough facetime with the doctor," said Michael Nash, a medical researcher and scientific advisor to Analytics Ventures.

Ali Parsa, Babylon’s founder and CEO, took a different view. “There's no way to make health care accessible and affordable if we have to send people to doctors, and doctors take as long as they currently do with each patient," he said. But the CEO agreed that in an ideal situation and in most developed countries, the best results would likely occur when doctors used the Babylon tool and in conjunction with their own skills and intuition. "They can improve each other," he said.

Babylon said its existing service – which lets people consult with a doctor over their mobile phone – has 2m users in Rwanda, a country where there are 0.064 doctors per 1,000 people, according to World Bank Data, compared to 2.8 in the UK and 2.3 in the US.

Babylon also pitted its software against a group of seven experienced primary care doctors on a set of 100 hypothetical cases developed by primary care experts at the Royal College of Physicians and the health systems at Stanford University and Yale University. Babylon’s software made the correct diagnosis in 80% of cases, while the human docs’ accuracy ranged from 64% to 94%, it said.

When the software was presented with a smaller set of cases that represented the conditions most commonly seen by GPs, its accuracy was 98%, the company said.

Business Report report

Receive Medical Brief's free weekly e-newsletter

Related Posts

Thank you for subscribing to MedicalBrief

MedicalBrief is Africa’s premier medical news and research weekly newsletter. MedicalBrief is published every Thursday and delivered free of charge by email to over 33 000 health professionals.

Please consider completing the form below. The information you supply is optional and will only be used to compile a demographic profile of our subscribers. Your personal details will never be shared with a third party.

Thank you for taking the time to complete the form.