Figure 4.

MS disease severity scale (MS-DSS) (at first clinic visit) is a strong predictor of future disability measured with therapy-adjusted CombiWISE slopes. (A) Cross-sectional MS-DSS measured at the first clinic visit strongly correlates with longitudinal therapy-adjusted CombiWISE regression slopes in the training cohort. (B) Relative influence of individual features in MS-DSS model. The Gradient Boosting Machine assigned importance to the following variables measured at the first clinic visit: therapy-adjusted CombiWISE/age, measured CombiWISE, COMRIS-CTD, time from disease onset to first therapy, difference between adjusted CombiWISE and unadjusted CombiWISE, age, and family history of MS. Known dichotomous classifiers of MS such as gender, race (Caucasian or non-Caucasian), and smoking history were determined to have little if any influence on model results. (C) Cross-sectional MS-DSS measured at the first clinic visit also strongly correlates with longitudinal therapy-adjusted CombiWISE regression slopes in the validation cohort. CombiWISE values were calculated based on a published formula (10), and subsequent therapy-adjusted CombiWISE slopes were derived using least squares fitting (see Materials and Methods) for patients with an average follow-up of 4.49 years.