Ankylosing spondylitis often takes a long time to be diagnosed via traditional methods like imaging, and as such, opportunities for early treatment intervention may be missed. With the condition affecting 1 in 400 people, a team of researchers from the United Kingdom sought to identify potential machine-learning methods to capture characteristics of patients who may be predisposed to developing ankylosing spondylitis in the future. In a paper published by Kennedy et al in PLOS One, they used data from the Secure Anonymised Information Linkage (SAIL) Databank based at Swansea University Medical School in Wales to compare data from patients diagnosed with ankylosing spondylitis, who were then matched with controls with no history of diagnosis. Data were analyzed by sex, and predictive models were developed; among men, lower back pain, uveitis, and receipt of nonsteroidal anti-inflammatory drugs before age 20 years were associated with ankylosing spondylitis, and for women, the researchers found an older age of symptom presentation than men with back pain and multiple pain relief medications. The model performed at a positive predictive value of 76.69% to 78.3% and was more accurate at prediction among women than men. The researchers brought up the possibility of applying multiple predictive models to improve prediction accuracy.
In a companion press release focused on the findings from Swansea University, study authors commented, “On average, it takes 8 years for patients with ankylosing spondylitis from having symptoms to receiving a diagnosis and getting treatment. Machine learning may provide a useful tool to reduce this delay…. Early detection and diagnosis are crucial to secure the best outcomes for patients. Machine learning can help with this. In addition, it can empower general practitioners, helping them detect and refer patients more effectively and efficiently.”