Predicting disease progression is essential to improve the health outcomes of patients with multiple sclerosis.
Multiple sclerosis (MS) progression may be identified through machine learning models, according to findings published by researchers in PLOS Digital Health. Predictive models measuring disease progression can be unreliable, calling for improved, advanced tools capable of accurately informing clinicians of disease progression in patients with MS.
The models were trained and validated using strict clinical guidelines to promote their applicability from research to practice. Image Credit: © tatomm - stock.adobe.com
MS is a neurodegenerative disease that affects approximately 1.8 million individuals around the world. It is characterized by the demyelination of the central nervous system that results in permanent damage of the myelin sheathes around nerve fibers, affecting cognitive, emotional, motor, sensory, and visual functions. MS is difficult to diagnose in its early stages and has limited treatment options, emphasizing the need for improved tools capable of earlier diagnosis and disease progression prediction to optimize outcomes for patients.1
Disability progression, especially with neurodegenerative disorders, is an essential milestone for determining disease evolution in patients with MS; however, there are few reliable prediction models or benchmarks to assess disease development and progression. To measure the probability of disability progression, state-of-the-art machine learning models were investigated by researchers from Belgium. They assessed 3-year follow-up data from 15,240 adults with MS from 146 MS centers in 40 countries. Using 2-year disease progression data from the study population, they trained machine learning models to predict disease progression over the subsequent months and years.2,3
The models were trained and validated using strict clinical guidelines to promote their applicability from research to practice. Area under the receiver operator curve (ROC-AUC), area under the precision recall curve (AUC-PR), and their calibration via the Brier score, were used to identify classification ability and precision and recall, respectively.2,3
According to their findings, the model performance across various patient subgroups had a ROC-AUC of 0.71 ± 0.01, an AUC-PR of 0.26 ± 0.02, and a Brier score of 0.1 ± 0.01. The researchers also reported an expected calibration error of 0.07 ± 0.04. Additionally, the researchers found that the history of disability progression was more predictive for future disability progression than treatment or relapse history.2
Models that accurately predict disability progression in MS are crucial and have the potential to aid medical professionals in treatment planning and treatment decision-making to enhance quality of care for patients. Further research is needed to determine the models’ applicability in clinical practice; however, the initial findings may pave the way for improved treatment of patients with MS.
“Using the clinical history of more than 15,000 people with [MS], we trained a machine learning model capable of reliably predicting the probability of disability progression in the next 2 years,” said study lead author Edward De Brouwer, MD, PhD, MEng, researcher at KU Leuven in Belgium, in a press release. “Our rigorous benchmarking and external validation support the vast potential of machine learning models for helping patients planning their lives and clinicians optimizing treatment strategies.”3
REFERENCES