Robust and Accurate Modeling Approaches for Migraine Per-Patient Prediction from Ambulatory Data.

摘要: Migraine is one of the most wide-spread neurological disorders, and its medical treatment represents a high percentage costs health systems. In some patients, characteristic symptoms that precede headache appear. However, they are nonspecific, their prediction horizon unknown pretty variable; hence, these almost useless for prediction, not useful to advance intake drugs be effective neutralize pain. To solve this problem, paper sets up realistic monitoring scenario where hemodynamic variables from real patients monitored in ambulatory conditions with wireless body sensor network (WBSN). The acquired data used evaluate predictive capabilities robustness against noise failures sensors several modeling approaches. obtained results encourage development per-patient models based on state-space (N4SID) capable providing average forecast windows 47 min low rate false positives.