BACKGROUND AND RATIONALE Magnetic Resonance Imaging (MRI) is the diagnostic cornerstone in Multiple Sclerosis (MS). Several MRI parameters are utilized in disease monitoring and in making treatment decisions. However, at the moment, baseline measurements of MRI parameters alone or even in combination with clinical parameters are insufficient to provide accurate predictions of risk for disease worsening over time. Being able to accurately predict an MS patient’s individual risk of worsening at baseline could help to personalize subsequent treatment decisions. Additionally, it would allow selection of more at risk patients for inclusion in clinical trials, allowing for both shorter duration of studies and smaller sample sizes, by minimizing the impact of inconclusive results from patients with stable disease. Recent advances in the field of machine learning and Artificial Intelligence (AI) pertaining to image classification have provided us with tools that can be applied on image data stemming from MRI. MRI imaging data can be analyzed using machine learning techniques such as deep neural networks to identify MRI image features accurately predicting the risk of disease worsening. STUDY OBJECTIVES PRIMARY OBJECTIVE To evaluate the ability of the machine learning model trained using MRI images, collected at baseline, from MS patients to predict disease activity in MS patients after 1 year. SECONDARY OBJECTIVE To assess features selected by the machine learning model with respect to classical clinical and MRI endpoints. STUDY DESIGN Patients with MS diagnosis, treated and followed per clinical praxis, at the Uppsala University Hospital, have been included in the SYMS (Synthetic MRI in Multiple Sclerosis) study. Their baseline and post-baseline (≥ 1 year) follow-up data will be analyzed in the current study. The data will be used to build and evaluate AI-based models for prediction of disease worsening. This data will include: • Demographics such as age, gender, disease duration, disease modifying treatments, time on treatment, working ability. • MRI imaging: synthetic MRI (QMR), morphological MRI (T2/FLAIR, T2 turbo spin echo, diffusion-weighted imaging, T1 before and after contrast agent injection), arterial spin labeling (ASL) perfusion (without contrast agent injection). • Clinical attributes, including, EDSS status, relapses, MSSS and transition to secondary progressive phase will also be collected at baseline and at least 1 year follow up or last post-baseline follow up. ENDPOINTS Primary endpoints Proportion of patients accurately predicted to have evidence of disease, i.e., being not NEDA-3, at least 1 year after baseline. NEDA-3 is defined as freedom from T1 Gd+ and new/enlarging T2 lesions, freedom from relapses and freedom from 3-month confirmed disability progression. Secondary endpoint Correlation of prediction features to MRI parameters (Brain parenchymal fraction [BPF], lesion count). Correlation of prediction features to clinical parameters: relapse rate, Expanded Disability Status Scale (EDSS) score, Multiple Sclerosis Severity Score (MSSS), conversion to Secondary Progressive MS.