New AI app may slash diagnosis time for people with EDS
HAT uses smartphone camera to analyze hypermobility
A new smartphone app, called the Hypermobility Assessment Tool (HAT), may aid in earlier and more accurate screening for Ehlers-Danlos syndrome (EDS).
Powered by artificial intelligence (AI), the app uses a smartphone camera to analyze how a person moves and identify hypermobile joints, a hallmark sign of EDS. Approved by Health Canada as a clinical screening tool, the app is meant to help patients and clinicians reach a diagnosis more quickly and accurately.
The app was developed by researchers at the University Health Network’s (UHN) Toronto General Hospital, in Canada. The team was led by Nimish Mittal, MD, co-medical director for the EDS Program at UHN’s Toronto General Hospital, and Babak Taati, PhD, a senior scientist at UHN’s KITE Research Institute. The researchers say the project was driven by a surge in patient referrals that followed greater EDS awareness.
“Because it’s a rare disorder, people with EDS are often told their symptoms are from fibromyalgia, or that it’s all in their head,” Mittal said in a UHN press release. “By putting an easy-to-use hypermobility assessment tool in the hands of clinicians and patients, we hope to empower both physicians, who don’t have enough knowledge about the disease, and patients who are concerned that they might have this disorder.”
How doctors currently assess hypermobility in EDS
EDS is an umbrella term for disorders that affect the connective tissue, which provides structure to joints, skin, blood vessels, and other tissues and organs. People with EDS typically have overly flexible joints and fragile, unusually stretchy skin, among other symptoms.
Healthcare providers currently assess hypermobility using the Beighton score test, which measures how far a person’s joints move during specific movements, focusing on the thumbs, elbows, knees, pinky fingers, and spine. However, the assessment may be imprecise, as it often relies on visual inspection or basic tools.
“Family doctors may not even know this test exists, may not have the device to measure the angles correctly, or may not know how to do it,” Mittal said. “This solves all those problems by automating the screening and the scoring, and it gives more power back to people concerned they might have this disorder to screen themselves.”
The HAT app guides users through nine standard movements, uses AI to analyze motion, and then generates a Beighton score. It can be used by clinicians and patients without any formal training.
Taati oversaw the app’s development, working closely with research associate Andrea Sabo. The team initially evaluated existing pose-tracking technologies but found they lacked precision when assessing joints that move beyond the normal range of motion.
“That led us to train and validate our own pose tracking models instead,” Taati said.
The team trained the new AI model on data from about 100 people, then tested it in 125 people referred to Mittal’s clinic for suspected hypermobility.
The study, which has been submitted for publication, found the app correctly identified more than 90% of positive cases.
The researchers are now comparing the app’s Beighton scores with results collected by healthcare providers in clinical settings to further test accuracy.
“Having a new, highly accessible screening option for EDS will benefit both patients and health care providers,” Mittal said. “We hope this app will reduce unnecessary anxiety and referrals for those who do not have the condition and shorten the long path to diagnosis for those who do.”
The free HAT app can be downloaded for Android and Apple devices.
About the Author
