Genetic disorders identified by Canadian-developed algorithm

Researchers at Children's Hospital of Eastern Ontario (CHEO) have developed the world's first artificial intelligence-powered algorithm to identify children with undiagnosed rare genetic diseases.

In a pilot project, researchers at the Ottawa Children's Hospital used AI to scrutinize approximately 260,000 patient records from the past 10 years. Analysis revealed that 15 of her children and adolescents likely had undiagnosed cases. Their symptoms were not amenable to genetic analysis.

There are approximately 6,000 rare genetic disorders, some of which are so rare that they occur in only one in five million people.

These illnesses are often complex and manifest as symptoms of other medical problems, leading to misdiagnosis and years of delays in treatment and care.

CHEO's ThinkRare algorithm, which took three years to develop, identifies likely cases much faster.

“Our research was focused on, 'How do we identify children who may have rare genetic diseases early in the diagnostic pathway to avoid the frustration that occurs later in life when they are ping-ponged around the system?'” said Kim Boycott, PhD, senior research scientist at the CHEO Institute.

Boycott said that if Sinclair had not identified these cases, some children may not have been referred to CHEO's genetic disease clinic for treatment.

George Whistaff's son was one of the cases identified by the algorithm.

Wistaff remembers getting a call from CHEO late last year asking her 10-year-old son, Anthony, to undergo genetic testing. His son was born prematurely and had undergone several surgeries to install shunts to treat hydrocephalus, a neurological disorder caused by an abnormal buildup of cerebrospinal fluid.

Anthony Wistaff, 10, and his parents Georges and Kimberly Wistaff. Anthony was diagnosed with a rare disease called Chung-Janssen syndrome. (Nice Peter)

As Anthony grew older, Whistaff said she noticed developmental delays in her son, including poor balance and difficulty retaining “everyday” information, but doctors couldn't explain why.

Over the course of 10 years, Anthony has visited CHEO over 100 times in six different specialty clinics, as well as the emergency department 30 times.

Then last August, the ThinkRare algorithm warned Anthony that he might have an undiagnosed rare genetic disease, and he was referred for genome sequencing testing, which analyses more than 5,000 genes linked to rare diseases.

In February 2024, Anthony was diagnosed with Chun-Janssen syndrome, a rare disease that has only been diagnosed in around 400 people worldwide. This syndrome explained many of Antony's health and developmental challenges.

“Anthony is actually the first person in North America to be diagnosed with this disease,” Whistaff said.

“One drop of blood answered so many questions.”

Wistaff said he and his wife can “stop pointing the finger” and focus on getting their son into treatment and programs to help him.

Dr. Boycott is also the director of genetics at CHEO, and she said a clinical diagnosis is not only key to receiving targeted treatment, it can also unlock valuable resources for families, such as access to special education assistants in the classroom.

“Diagnosis opens doors for not only the health care system but also the community. When school systems don't have a specific diagnosis code, they often don't have enough resources.”

Ivan Terekhov, CHEO's director of research informatics and technology, helped create the ThinkRare algorithm. He met the Whistaff family and called it an “amazing” and moving experience: it was the first time he saw his tech skills directly impact patient care.

Telekov said ThinkRare shows how AI can positively impact the medical field.

“AI is not replacing doctors. We are supporting them with digital tools. I don't think this is scary. I think it's exciting, but we need to use it responsibly.” need to do it.”

Until now, the ThinkRare algorithm has been used to identify historical cases that may have been missed, but it will soon be used to scrutinize the records of all current and new patients at CHEO.

Although CHEO could be the first hospital to use this general-purpose AI diagnostic tool, it has no plans to monetize the algorithm. Instead, Boycott said her team wants to distribute it for free so that pediatric hospitals around the world can more quickly identify genetic diseases in children.