🇬🇧 Artificial intelligence in rheumatology: an innovative approach
Nov 10, 2020 | 1:15 PM CET - 1:30 PM CET
Nov 10, 2020 | 1:15 PM CET - 1:30 PM CET
Description
Rheumatoid arthritis (RA) is an autoimmune disease that affects more than 20 million patients worldwide. Given the chronic nature of RA treatment, the identification of real-world predictive factors of treatment retention in patients with RA could assist rheumatologists with therapeutic decision-making in clinical practice and support a personalized approach to medicine. To identify predictors of 1-year retention on drug, multivariable Cox proportional hazards regression models were used to analyze data from a non-interventional multicenter prospective longitudinal clinical study of RA patients. To further understand the initial results from modeling, artificial intelligence (AI) was added as a complementary tool. In total, data from 2350 patients was included in the initial model and it was determined that the gradient boosting classifier model had the best prediction testing accuracy (67%) and was the most interpretable model for feature importance. The gradient-boosting model identified 51 predictors of retention, 8 of which overlapped with those identified by multivariable Cox regression models in the clinical. Machine learning offers a complementary approach to statistical modeling. The talk will discuss how use of AI may lead to more accurate identification of subsets of patients with RA who will be more likely to remain on therapy, hence supporting personalized, clinical decision making in a real-world setting. This session will be held by Ms. Karissa Lozenski, Ph.D who is the Director, Clinical Trial Lead, Medical Data Generation, Immunology & Fibrosis at Bristol Myers Squibb and Ms. Claire Behar, Director of the Data Science department at Excelya.
