Multiple Sclerosis (MS) is one of the most common neurological conditions worldwide whose prevalence is now greatest among people 50-60 years of age. While clinical presentations of MS are highly heterogeneous, mobility limitations are one of the most frequent symptoms. In contrast to the monitoring of most underlying manifestations of MS, which require neurological examinations by a trained practitioner, gait can be quickly and remotely monitored. In this work, we propose VGA4MS (vision-based gait analysis framework for MS prediction), a deep learning (DL) based methodology to classify gait strides of individuals with MS from healthy controls, so as to generalize across different walking tasks and subjects, on the basis of characteristic 3D joint key points extracted from multi-view digital camera videos. This is the first attempt to demonstrate the potential of vision-based DL for MS research. Given digital cameras are the only required equipment, this can be employed in the domestic environment of elderly for regular gait monitoring, and thus is crucial for early intervention and hence, more efficient MS treatment.