Shriners Children’s Scoliosis Research Looks for Breakthroughs with AI

On a chilly Chicago spring weekend, Alec’s Court at Shriners Children’s Chicago was transformed into a world-class research stage. Several of the presentations at this year’s State of Science conference focused on scoliosis, an abnormal curvature of the spine, which physicians across the Shriners Children's healthcare system treated in some 50,000 children over the past decade. Having a patient population this large, along with some of the largest groups of pediatric orthopedic surgeons in the country, is a recipe for treatment advances and research which may improve the care of scoliosis patients everywhere. As we celebrate Scoliosis Awareness Month in June, here are updates on the research they shared.

Can artificial intelligence (AI) provide more standardized and efficient measurement of spine curves?

Yishan Zhong, Ph.D., at the Georgia Institute of Technology, is working with spine surgeon Selena Poon, M.D., from Shriners Children's Southern California to develop an AI clinical tool for Cobb angle measurement. This measurement of the severity of a spinal deformity is determined by the physician marking lines on a spine X-ray above and below the largest part of a curve to see where they intersect. “Current manual scoliosis assessment is time-consuming for clinicians who are already overwhelmed. In addition, the subjective knowledge and experience of different clinicians result in variation in the measurement,” they wrote. They studied 1,000 scoliosis patients’ data to develop the AI model, which one day could be in patients’ medical records alongside other data and notes, to help inform treatment decisions. The study involves Shriners Children’s locations in Southern California, Lexington and Philadelphia.

Shriners Children’s Chief Medical Officer Fran Farley, M.D., commented after the talk, “Physicians diagnosing scoliosis really use the Cobb angle. This found that AI could help measure the curve. This shows lots of potential for minimizing time to make the measurement, but hopefully also making the measurement more accurate.”

Improving Care of Early Onset Scoliosis

Children who develop scoliosis before age 10 often face an issue with the curving spine taking up space needed in their chest for the lungs to mature. “Can we anticipate complications ahead of time using machine learning?” asked Steve Hwang, M.D., pediatric spine surgeon and neurosurgeon at Shriners Children’s Philadelphia. The new study in collaboration with Shriners Children's Chicago and Georgia Tech began by looking at genetic results from a population in the UK. The next steps will involve data from Shriners Children’s patients. They hope to identify genetic markers to guide future treatment decisions in these younger children with scoliosis.

Predicting Complications Prior to Surgery

Moving into the surgical area of care, Bruce Brenn, M.D., chief of anesthesia at Shriners Children's Philadelphia, is managing a study called FUSION to further protect children’s spinal cords in the operating room during spinal fusion surgery. Anesthesiologists carefully monitor interoperative neuro alerts which could indicate spinal cord distress for a patient, a very rare but potentially serious complication. The study will use AI algorithms to examine the large spinal fusion population across our system to develop a risk score for patients about spinal cord distress, using operating notes of previous cases and the large amount of data collected prior to surgeries.

Predicting Growth and Curve Progression in Idiopathic Scoliosis

Michelle Welborn, M.D., of Shriners Children’s Portland, provided updates on her groundbreaking work with the collagen X biomarker. She shared that in studies, this biomarker – first identified by William Horton, M.D., former director of research at Shriners Children’s Portland, and his team – was the strongest predictor of change in curve progression. “If those biomarker levels were spiking, those were the kids who change for good or bad,” she said.

This tool could one day help physicians better predict when to stop brace wearing in a child with scoliosis. Currently, the decision is tricky because doctors have to assess when a child is finished growing based on a clinical assessment. Doctors would like to further reduce unnecessary brace wearing and better identify those who should continue.

In order to develop a new risk severity score for idiopathic scoliosis patients, Dr. Welborn’s ongoing research will use AI to combine data from biologic, genomic and radiographic measures to predict growth patterns, skeletal maturity and likely curve progression.

These presentations are summarized from State of Science with a focus on muscle and bone health, a conference organized annually by Shriners Children's Chicago research staff to increase collaboration, share updates and identify new avenues of research which may one day improve the care of children everywhere with orthopedic conditions.