From behavioral to structural correlates: a multi-phase analysis of global reported mpox

Area of Science:

• Epidemiology
• Public Health
• Machine Learning

Background:

• Mpox (mpox) epidemiology evolved from zoonotic spillover to sustained human transmission.
• Macro-level correlates of mpox burden across epidemic phases are not well understood, hindering targeted responses.

Purpose of the Study:

• To quantify macro-level correlates of mpox burden across distinct epidemic phases.
• To develop a phase-sensitive analytical framework for understanding mpox epidemiology.

Main Methods:

• Utilized World Health Organization (WHO) global daily mpox surveillance data (Jan 2022 - Apr 2025).
• Calculated country-period mpox incidence and employed a phase-sensitive framework with ensemble machine learning models.
• Interpreted model outputs using SHapley Additive exPlanations (SHAP) values to identify changing correlates.

Main Results:

• Identified four temporal-epidemiological phases of mpox: emergence, spread, dissemination, and resurgence.
• Revealed a shift in correlates from behavioral factors (e.g., close contacts, mobility) to structural vulnerabilities (socioeconomic disparities, health capacity).
• The LGBT+ Legal Index was the strongest predictor, with context-dependent associations influencing case ascertainment and reporting.

Conclusions:

• Global mpox burden transitions from behavioral to structural indicators, showing phase-specific and geographic heterogeneity.
• Supports phase-tailored interventions combining prevention, health-system strengthening, equitable surveillance, and stigma reduction.