Table 3.

Main articles using AI for migraine attack forecasting.

Reference (First Author, Year)	Type of AI	Main Results	Number of Patients/Data Analysed
Katsuki, 2023 [49]	Statistical and deep learning-based methods	Low barometric pressure, barometric pressure changes, high humidity, and rainfall were associated with an increased occurrence of headache attacks	4375 filtered users with 336,951 headache events
Stubberud, 2023 [48]	Several standard ML architectures with random forest classification show the best performance	An AI migraine attack forecasting model is possible; the most predictive factors were premonitory symptoms of craving, swelling, feeling cold, the amount of sleep, and the presence and intensity of headache	18 patients who completed 388 headache diary entries

Abbreviations: AI = artificial intelligence; ML = machine learning.