White Coat Black Art26:30The power of AI to diagnose rare diseases
If it hadn’t been for a Google search 10 years ago, Ian Stedman may never have discovered he had a rare disease and, most importantly, wouldn’t have known how to treat it.
Now he hopes the growing development of artificial intelligence will make sure others don’t go undiagnosed for decades, like he did.
“I think [artificial intelligence] has potential to completely transform the health-care system,” Stedman told White Coat, Black Art host Dr. Brian Goldman.
“I don’t know if it’s for better or for worse yet, but I do think it’s a real important conversation for us to have about how to regulate it and how to make sure that what people are finding online is helpful instead of harmful.”
Stedman, of Woodbridge, Ont., went 32 years of his life without being diagnosed. But by plugging his symptoms into an online search, he figured out that both he and his daughter might have a rare disease called Muckle-Wells syndrome.
A new program at CHEO, eastern Ontario’s children’s hospital in Ottawa, that harnesses the power of AI may have been able to detect the disease a lot faster.
It says it’s the first Canadian hospital to use AI to assist in diagnosing rare diseases, and Stedman and researchers at CHEO hope there are many more programs like it to come.
Ian Stedman’s story
Ian Stedman, 43, grew up suffering from a skin rash, red eyes, migraines, arthritis and eventually partial hearing loss.
Despite doctor visit after doctor visit, there was no diagnosis. Stedman estimates he saw dozens of physicians during nearly 200 visits. He missed school and suffered at work. He never wore short sleeves because of his rash. He just lived with it.
But the birth of his daughter, Lia, who started exhibiting similar symptoms, made the situation a lot more serious. More doctor visits, more conversations with specialists and even reviewing medical journals still left Stedman without answers.
So he turned to Google — and after browsing thousands of photos, he found skin that looked like his, connected to Muckle-Wells syndrome.
After getting the diagnosis confirmed by Dr. Ronald Laxer, a pediatric rheumatologist at the Hospital for Sick Children in Toronto, Stedman’s life changed. He was able to get treatment for himself and Lia. Every two months, Stedman and his daughter take medication through a syringe that keeps their symptoms at bay.
It means Lia, 12, hasn’t had to experience those same symptoms the way he did.
“If you asked me what are the symptoms of Muckle-Wells, I used to be able to rhyme them off,” Stedman said.
“Ten years later, it’s not as easy for me to just rhyme them all off…. I have to actually sit and think because I’m so far removed from having to experience them.
The power of AI
After his diagnosis, Stedman joined the board of the Canadian Organization for Rare Disorders, serving for a three-year term. He wanted to become an advocate for those like him who are living with rare diseases. More than that, he didn’t want those diseases to continue to go undiagnosed.
He said while doctors are smart, it’s impossible for them to know everything.
“That’s why the system has to find a way to be more intelligent, to support physicians,” Stedman said.
That’s happening through programs such as ThinkRare, where researchers at CHEO have developed an algorithm to help identify rare genetic diseases in children.
It takes the information stored in a patient’s electronic health record and cross-references the different departments where the child has been seen.
“So we’re looking for complex children in the hospital who have multi-system involvement, who have not yet been assessed for a rare genetic disease,” said Dr. Kym Boycott, a professor of pediatrics at the University of Ottawa and a clinician scientist at the CHEO Research Institute.
When the system detects a potential rare genetic disease, it flags it to Boycott’s team, and contact is made with the patient’s primary physician.
So far the system has flagged about 250 patients who could have a rare genetic disease, and of those, 50 have been referred for genetic assessment. Those on the team have been able to test 19 children, and while some results are still pending, they’ve been able to identify seven patients with genetic diseases who are now receiving treatment.
“The ultimate goal was to use AI to bring the diagnostic genetic testing to the front of the care pathway and not at the back,” Boycott said. “It’s about catching kids early, not catching kids that we’ve missed.”
Alexandre White-Brown, a research co-ordinator with the ThinkRare project, said it’s connected with other organizations across Canada about how they could incorporate a similar algorithm of their own to help diagnose rare diseases.
“Our goal isn’t to make money off this or to sell this or to patent this. Our goal is really to share this to allow for rare disease diagnoses across Canada,” White-Brown said.
“This can change people’s lives. It can save lives.”
A new Canadian study found that an artificial intelligence early warning app helped prevent unexpected hospital deaths by 26 per cent. The technology isn’t meant to replace medical staff, but serve as an additional tool for patient care.
Security concerns
Detecting rare diseases is far from the only way AI is being utilized in the medical field.
Doctors have started using AI to transcribe and summarize conversations with patients. AI-based early warning systems for patients in hospitals have been found to dramatically decrease the number of unexpected deaths, according to a study published in the Canadian Medical Association Journal.
But as artificial intelligence becomes ubiquitous in society and hospitals, some experts are calling for caution.
“I don’t think AI in and of itself is the issue,” Dr. Sheryl Spithoff said. “I think there are still issues of understanding how this is going to be incorporated into practice.”
Spithoff is a scientist at the Women’s College Research Institute and an assistant professor at the University of Toronto’s department of family and community medicine.
She researches how AI is used in the medical field — and specifically who funds these programs in Canada. She’s found that often pharmaceutical companies are sponsoring IT companies to create algorithms.
“[The systems] are then largely used to identify patients … who may be appropriate for treatment with the pharmaceutical companies’ drug product, which is usually an expensive new or patented drug,” Spithoff said.
The first step with any AI program used in the medical field is making sure there is transparency around who is funding it, she said, adding there needs to be more public funding, so the programs aren’t being made by companies driven by profit.
Spithoff said ThinkRare at CHEO, which was funded by donations to the CHEO Foundation, is a good example of a positive way to harness the power of artificial intelligence. Even still, hospitals and organizations that do this need to be hyper-aware of data privacy, she said.
“These are the tools that can potentially make a big difference, but it really has to be in the hands of patients and the public to make these decisions — of course, with the guidance of medical experts and researchers, but to figure out what their priorities are,” Spithoff said.
Ian Stedman said while he understands people’s trepidation, he believes that’s why the use of AI needs to be researched now.
“I don’t have misgivings about the technology’s potential. I have misgivings and worry that we are going to allow the technology and its creators to do what they want without the right safeguards in place,” he said.
Interviews with Ian Stedman and Kym Boycott poduced by Colleen Ross and Sameer Chhabra.