Braive is an online platform that helps people facing mental health challenges, to learn strategies for building skills required to cope with stress, anxiety, etc. People with mild-symptoms can complete psycho-education based programs, to avoid longer queues in hospital; or indicative diagnosis to later discuss with general practitioner. In this paper, we apply different ML approaches for Success Prediction to estimate the outcome of treatment, given initial mental health assessment, enrolled programs, and condition monitoring test scores. An extensive experimentation of the ML approaches, using feature ablation settings, show improved prediction performance, when the time span of treatment is increased.