On Claire Novorol’s first day as a geneticist, she encountered an unknown illness. A baby was brought into Addenbrooke’s Hospital in Cambridge, suffering from symptoms that matched no known disease. Novorol was shocked. “I had no background in this area and, looking at the notes, I didn’t know what this was,” she says. The baby had been seeing doctors for over a year.
By combining information from databases, symptoms and case studies, Novorol found the correct test, which led to a diagnosis. (To this day, the case is so rare that she cannot name the disease, as it risks identifying the patient.) But that episode taught her the limitations of humans, especially when it comes to retaining information. “The knowledge just doesn’t fit inside the heads of doctors,” she says.
Novorol, 39, is co-founder of Ada, a Berlin- and London-based startup whose app gives its users information about their symptoms. Answer a series of questions, and you’ll get guidance about your ailment, including an assessment of what it might be. The app, which is available on Android and iOS, can’t give a formal diagnosis – there are strict regulations around diagnosing patients – but users can share their results with their doctor. Ada is also planning on allowing people to connect people with chosen healthcare partners. Since launching in the UK in April 2017, Ada has been downloaded by three million people worldwide.
Ada hasn’t always worked this way. When the company launched in 2011, co-founders Daniel Nathrath and Martin Hirsch set out to help perplexed doctors diagnose rare and complex conditions. “They were focused on vertigo and then looking to extend out to neurology,” says Novorol, who joined a few months after the company launched. The problem: overstretched doctors didn’t have enough time to enter the symptoms and medical histories of complex patient cases into yet another system. So instead, Ada shifted its focus toward patients, building a medical database to match symptoms and ailments.
“We moved from one speciality to multiple specialities, then to covering all of general practice,” explains Novorol. Ada’s database now contains more than 1,500 conditions, she says, based on over 5,000 findings from scientific studies. But even once it had all this information, Ada was left with a challenge. To draw information out of humans, the app needed to talk to them – so Novorol and the team started to develop a conversational artificial intelligence engine.
On launching the app, Ada’s “patients” are presented with a long list of questions based on the symptoms they have entered (“Have you lost your appetite?”, “Do you have any lumps under your abdominal skin?”). For each question – a typical test can involve more than 20 – the app presents an explanation of any medical complexities. Novorol says Ada’s AI has been built to be “friendly, but at the same time authoritative”. It’s not a doctor, but has the air of one.
Several startups, such as babylon and Your.MD, are attempting something similar – and the stakes are high for the company that comes out on top. According to the World Health Organization’s 2017 Tracking Universal Health Coverage report, more than half of the world’s population doesn’t have access to health services. In India, poor access to public healthcare forces people to turn to private services or forego basic analysis. (Ada has “several hundred thousand” users there, Novorol says). At present, lots of people turn to “Doctor Google” – but dedicated apps do a better job at symptom checking, according to a study published by the British Medical Journal in 2015.
Read more: Can you really trust the medical apps on your phone?
So how does Ada stack up against its competitors? “The questions are clear, and it translates free text into sensible suggestions for the user to choose,” says Hamish Fraser, senior lecturer in eHealth at the University of Leeds. In our symptom-checking-apps test, Ada’s deep knowledge and conversational abilities made it the clear winner. “Ada was by far the best,” says David Wong, lecturer in health informatics at the University of Leeds. “There were issues with the others on test. It was surprising to be able to find things wrong in a few minutes, from a non-clinical perspective.”
In October 2017, Ada raised €40 million (£35.2m) in funding and plans to use the money to open a US office. But Novorol’s long-term ambition is to be more involved in people’s overall health. She says most of us are generally aware of the need to exercise regularly and eat healthily – we just find it difficult to act on that knowledge. But interrupting a person’s routine with advice that’s based on their medical history may be the answer. “The next step is really to tell people what to do,” Novorol adds. “There’s a lot of great knowledge and theory on how to nudge someone in the right direction.”
Updated: April 23, 2018: This piece has been amended to include new user figures and remove a reference to video conversations.
This article was originally published by WIRED UK
