When will patients see an AI doctor?

How would you feel if your doctor was a robot? Uncomfortable? A little wary? Or maybe reassured because you’ve lost faith in the real thing?

Well, don’t panic because you won’t be seeing AIs taking over your local surgery or district hospital for a few years at least and if they do, they won’t be left in charge.

“We’re a good way from AIs directly treating the public but we will soon start to see robots in certain capacities working alongside humans – what we call ‘human-in-the-loop’ collaborations,” explained Professor Dhiya Al-Jumeily, a computer scientist at LJMU.

Dhiya is training his own ‘robot doctor’ with research funding from LJMU and based on a 25-year career in machine learning and healthcare applications. To date, Ami can already provide more accurate predictions than human doctors and quicker and there’s a lot more to come with further upgrades.

Ami is a piece of hardware, created by Engineered Arts, UK, and equipped with a large language model that enables human-like speech. What sets her apart from other robots and makes her suited to healthcare interactions is that Ami is capable of warm, friendly and almost human conversation. She can simulate emotions via facial expressions that express happiness, sadness, anger, fear, surprise, worry; something the engineers think is a key to her success.

“Accepting technology is a big leap, especially when it comes to our own health and wellbeing,” says Dhiya. “The best way around it is breeding familiarity, so we are doing a lot of work to bring Ami into the healthcare world and break some of the boundaries between humans and humanoids.

Ami is essentially being taught, not to be a doctor per se, but to synthesise scientific information and disseminate it to people. She has trained to observe and question not only data but the knowledge acquired from more than 40 doctors involved from the NHS and the World Health Organisation.

In October, Dhiya and his team introduced Ami to medics in London and ‘she’ took part in a talk about AI past, present and future, answering questions from healthcare professionals from different disciplines.

In the current Industry 4.0 climate, many organisations are exploring the use of AI bots, in receptions or call centres to offer a ‘digital front door’ to public services. Dhiya sees a much more advanced role for the likes of Ami, in a world where training more human doctors is prohibitively expensive. 

“We’re not far from having 10 billion people on Earth and not nearly enough doctors to serve them,” adds Professor Al-Jumeily, who explains that if trained correctly and with the right data sources, that within five years, the likes of Ami will be offering highly accurate diagnoses, reducing medical error and improving treatment outcomes.

“I think we’ll certainly see them working alongside humans offering accurate and fast decision-making.”

But the LJMU scientists also recognise the limits, noting that from interactions to date, that people’s reactions to Ami are mixed just like reactions to any new technology.

“Some individuals do welcome contribution from technology and others show fear of working alongside robots. We are trying to address this fear by involving doctors in workshops, conferences and presentations with Ami and also involving them in their own medical training.

“At the end of the day, it’s about what is useful but also what people will accept. If people want to see a real human doctor, there’s no point arguing that the robot knows better.” “I can’t see robots either replacing humans or even taking major medical decisions.”

It’s also a team effort from LJMU: “The School of Computer Science and Mathematics and the Institute of Healthcare Research have been supporting us in applications and in linking us with practitioners from the NHS that are co-applicants in our research funding applications.”

One such bid is now in the pipeline with Imperial College and the World Health Organisation involves Ami and developing training courses for healthcare practitioners in artificial intelligence and medical datasets.