One of the most disabling neurological disorders among patients is migraine. Its prognosis and outcome can be in influenced by different factors, for this reason, there are a lot of studies to investigate the causes of this type of disease. The aim of this work is to highlight the relation between early traumas, stressful life events and the outcome of Medication Overuse Headache patients after a one-year detoxification therapy. A two-step approach is proposed, applying both unsupervised and supervised machine learning methods to evaluate the multivariate relation among static information, collected at the baseline from surveys, and patients’ individual trajectories, which represent the evolution of migraine status in a one-year study. All the proposed analysis and the presented results are based on a real dataset. The final outcome of this study demonstrates that it is possible to identify different reactions of the patients under treatment. Therefore, using the patients’ information given at the baseline, prediction models can be successfully employed to explain different responses in advance with the aim of paying more attention to the most critical patients.

sequence clustering, classification models, migraine, life events.