Predicting Individual Outcomes in People with Multiple Sclerosis

Multiple sclerosis is an incurable chronic inflammatory disease that typically first manifests in early adulthood and causes accumulating disability through damage to the central nervous system. Both disease course and response to current treatment options vary across individuals, making it difficult to make accurate prognoses and treatment decisions at the individual level. Clinical prediction models, formal multivariable combinations of predictors that yield probabilities of disease states or outcomes, may be used to assist in these tasks, among others. This thesis comprises a Cochrane systematic review which aimed to identify, describe, and assess the current state of the MS prediction modeling literature, with a focus on prognostic models for disease progression, worsening, and activity, in order to inform further MS prediction research, including validation of promising models and development of novel models, where needed. The novelty of this review also allowed for a description of current gaps in the methodological guidance regarding assessment of models developed using complex machine learning methods somewhat beyond the scope of the current guidance