Date of Award

January 2023

Document Type

Open Access Thesis

Degree Name

Master of Public Health (MPH)


School of Public Health

First Advisor

Judith Lichtman


Background – Headache is one of the significant public health threats in the world. Patients often face difficulties preparing for medications before headache attacks due to their unexpected nature. A headache diary was previously employed in studies to develop a machine-learning prediction model. However, previous research used variables related to a narrow range of topics or used ensemble methods where the decision trees were not added sequentially. Therefore, this study aimed to use a gradient-boosting approach to find factors from a headache diary with a broader range of factors with strong predictive values for headache prediction.Methods – The three-month headache diary data were collected thrice daily using the Status/Post Apple device application. A total of 23 adult patients’ self-reported data was used in a gradient-boosted classification machine-learning model. The primary outcome measure was the self-reported absence/presence of a headache, and the features used were self-reported symptoms and lifestyle variables asked in a headache diary. Results – The gradient-boosting classifier model’s area under the curve (AUC) for 23 adult headache patients was 0.94, which shows a strong differentiating ability between the probability of having a headache and not having a headache. Also, the model’s ability to accurately predict headaches was 0.80, shown by the F1 score of the model. The study also discovered that premonitory symptom variables were more predictive than others. Conclusion – This study shows that future headache attacks can be accurately predicted for adult headache patients using the GBM classification model. Additional research is needed to explore whether the model can be used in other populations and whether a strong predictive model with fewer and stronger predictive variables found in this study can be developed. Keywords: Chronic Disease Epidemiology, Headache, Migraine, Prediction, Machine Learning


This is an Open Access Thesis.

Open Access

This Article is Open Access