AHS 2022 | Challenges in using AI deep learning in migraine research

Mahfuzur Rahman Siddiquee, MD, Arizona State University, Phoenix, AZ, discusses the main challenges concerning artificial intelligence (AI) in migraine classification. AI deep learning is dependent on good data sets; thus, one of the major challenges in this field is having enough data. With only 67 migraine patients in the study and approximately 500 healthy patients, there was a significant data imbalance – leading to the AI attempting to predict everything as healthy. Furthermore, due to the nature of migraine, patients were in more discomfort, thus moving more during the MRI scan and leading to the blurring of certain regions. This interview took place during the 2022 American Headache Society (AHS) Annual Scientific Meeting in Denver, CO.