Driving Mechanisms and Future Trends of Migraine and Tension-Type Headache Burden in G20 Countries: A Comparative Analysis Utilizing Decomposition Analysis and Explainable Machine Learning.

Abstract

Introduction: Migraine and tension-type headache (TTH) are leading causes of neurological disability globally. While their macro-trends are documented, the nonlinear drivers and the regulatory role of socio-economic development across G20 countries remain under-explored.

Methods: Based on Global Burden of Disease (GBD) 2023 data, the headache burden across G20 members was quantified. Developmental trajectories were identified by integrating the Socio-demographic Index (SDI). Crucially, a comparative analysis was conducted between a decomposition analysis model and an explainable machine learning framework, Shapley Additive Explanations-Extreme Gradient Boosting (SHAP-XGBoost), to quantify nonlinear marginal contributions and complex feature interactions. Finally, future trends from 2024 to 2035 were projected utilizing Autoregressive Integrated Moving Average (ARIMA) models.

Results: In 2023, the number of prevalent cases for migraine and TTH across G20 countries was estimated at 705 million and 1.27 billion, respectively. A significant nonlinear correlation between headache burden and the SDI was identified, with the burden for both disorders peaking at an SDI of approximately 0.8 before plateauing or declining. While population growth was identified as the primary macro-driver in both decomposition and SHAP analyses, a "saturation effect" within the headache models was revealed by SHAP analysis. Specifically, the marginal contribution of population size was found to diminish once specific thresholds were exceeded. Furthermore, according to ARIMA model projections, the age-standardized incidence rate (ASIR) of migraine is expected to stabilize in the coming years, whereas the ASIR of TTH is forecast to maintain a continuous upward trajectory through 2035.

Conclusions: This study provides a comprehensive re-evaluation of migraine and TTH burden across G20 countries, shifting the focus from descriptive status quo to mechanistic explanation. Our findings confirm that while population growth remains a key contributor, variations in burden appear to be associated with differences in development level, age structure, and population scale.