Personalized machine learning prediction of next-day migraine persistence using digital headache diary data.

Migraine is a complex neurological disorder with considerable individual differences. Predicting if a migraine will continue into the next day is difficult. This study assessed personalized machine learning models for predicting next-day migraine persistence. Researchers used longitudinal digital headache diary data for their predictions. Participants were recruited from two medical centers in Taiwan from February 2023 to October 2023. Each participant completed a three-month headache diary observation period. Patients with five to fourteen headache days each month were included. The study employed models like k-nearest neighbors, support vector machines, random forest, and eXtreme Gradient Boosting. The personalized KNN model showed significantly better performance than the generalized version. It achieved an area under the curve of 0.83 compared to 0.63. The study suggests that personalized models can effectively predict migraine persistence. Further research is necessary to confirm these findings and their clinical relevance. Clinical relevance: This research indicates that personalized machine learning can improve migraine prediction accuracy, potentially enhancing patient management strategies.