Houston lab develops AI tool to improve neurodevelopmental diagnoses
developing news
One of the hardest parts of any medical condition is waiting for answers. Speeding up an accurate diagnosis can be a doctor’s greatest mercy to a family. A team at Baylor College of Medicine has created technology that may do exactly that.
Led by Dr. Ryan S. Dhindsa, assistant professor of pathology and immunology at Baylor and principal investigator at the Jan and Dan Duncan Neurological Research Institute at Texas Children’s Hospital, the scientists have developed an artificial intelligence-based approach that will help doctors to identify genes tied to neurodevelopmental disorders. Their research was recently published the American Journal of Human Genetics.
According to its website, Dhindsa Lab uses “human genomics, human stem cell models, and computational biology to advance precision medicine.” The diagnoses that stem from the new computational tool could include specific types of autism spectrum disorder, epilepsy and developmental delay, disorders that often don’t come with a genetic diagnosis.
“Although researchers have made major strides identifying different genes associated with neurodevelopmental disorders, many patients with these conditions still do not receive a genetic diagnosis, indicating that there are many more genes waiting to be discovered,” Dhindsa said in a news release.
Typically, scientists must sequence the genes of many people with a diagnosis, as well as people not affected by the disorder, to find new genes associated with a particular disease or disorder. That takes time, money, and a little bit of luck. AI minimizes the need for all three, explains Dhindsa: “We used AI to find patterns among genes already linked to neurodevelopmental diseases and predict additional genes that might also be involved in these disorders.”
The models, made using patterns expressed at the single-cell level, are augmented with north of 300 additional biological features, including data on how intolerant genes are to mutations, whether they interact with other known disease-associated genes, and their functional roles in different biological pathways.
Dhindsa says that these models have exceptionally high predictive value.
“Top-ranked genes were up to two-fold or six-fold, depending on the mode of inheritance, more enriched for high-confidence neurodevelopmental disorder risk genes compared to genic intolerance metrics alone,” he said in the release. “Additionally, some top-ranking genes were 45 to 500 times more likely to be supported by the literature than lower-ranking genes.”
That means that the models may actually validate genes that haven’t yet been proven to be involved in neurodevelopmental conditions. Gene discovery done with the help of AI could possibly become the new normal for families seeking answers beyond umbrella terms like “autism spectrum disorder.”
“We hope that our models will accelerate gene discovery and patient diagnoses, and future studies will assess this possibility,” Dhindsa added.
