G. Van Gils
Aim: To develop and evaluate a contactless, camera-only method for measuring pre-attack physiology in migraine at home, and to assess its feasibility over four weeks. Methods: In this longitudinal observational feasibility study, adults with a neurologist diagnosis of migraine used a browser-based, camera-only application on their own device three times daily for four weeks. The application derived heart rate and heart-rate variability from facial video, together with proxies for eyelid and mouth movement and blink rate, each referenced to a person-specific baseline. Measurements within twelve hours before a self-reported attack were compared with attack-free measurements using linear mixed models, and individual prediction was tested with leave-one-subject-out cross-validation. The complete analysis was independently re-implemented and reproduced. Results: Nine of thirteen recruited participants completed the study and contributed 549 measurements (547 reliable). Heart rate was higher in the twelve hours before an attack than in attack-free periods (66.0 versus 60.4 beats per minute), confirmed by a mixed model (plus 5.54 beats per minute) and unchanged after adjustment for stress; no other signal differed after correction. Individual-level prediction remained weak (best cross-validated area under the curve 0.58). Conclusion: Contactless, camera-only measurement of pre-attack physiology at home is feasible and reproducible. A group-level rise in heart rate was demonstrable but did not translate into usable individual prediction, motivating a larger study with electrocardiographic validation.