The associate professor of neurology at Columbia University and medical consultant and care center director at the Muscular Dystrophy Association discussed the use of larger datasets to improve prognostication and clinical trials in ALS.
“Maybe 5 or 6 years ago, there was a publication that took the aggregated clinical data from 1200 patients, and basically asked, ‘Can machine learning pick out different subgroups and prognosticate how those groups are going to do?’ That looked pretty good, and it definitely allowed us to get a sense for people’s progression—but the error bars were huge. A couple of years ago, now, that number has grown to 12,000…and the error bars shrunk considerably.”
Big data has developed into a hot button topic in the medical field, with advances in collection, registry growth, and machine learning allowing for the compiling of massive databases that can better inform prognostication and treatment decisions. For some disease states, these databases can hold information from individuals numbered in the millions. But for others, like amyotrophic lateral sclerosis (ALS), those numbers are closer to the thousands.
Even still, these advances have made an almost immediate impact for physicians who are caring for these patients. Matthew B. Harms, MD, associate professor of neurology, Columbia University, and medical consultant and care center director, Muscular Dystrophy Association, told NeurologyLive that while statisticians may scoff at the idea of these datasets being referred to as “big data,” they’ve made a tangible difference in the care of patients with rare diseases like ALS.
Harms shared his experience with these advances and the effect it has had on not only prognostication, but on clinical trial stratification and design. These datasets, he said, have allowed for more easily interpretable clinical trial data, and have laid the groundwork for the push toward precision medicine for this population.