Deterioration of immunization resilience: A Bayesian machine learning assessment of the 2025-2026 measles outbreak in Mexico

OBJECTIVE: The 2025 measles outbreak in Mexico (5741 cases) marked a severe decline in immunization resilience. We aimed to measure this systemic vulnerability and project 2026 coverage. DESIGN OR METHODS: A Bayesian Ridge Regression model was built using a high-resolution (2015-2025) dataset (WHO indicator WHS8_110 and national bulletins). To validate the model, Leave-One-Out Cross Validation and Bootstrap (n = 5000) methods were used. RESULTS: Our results project a MCV1 vaccine coverage of 85.41% (95% confidence interval: 64.45-100.00%) for 2026, which remains below the 95% herd immunity threshold. High postpandemic volatility limited the model's predictive performance (R² = -0.36), reflecting an atypical deviation from historical trends. Prior coverage (89.59%) is the primary driver, while outbreak load and crisis status (10.41% combined) trigger the projected downturn. CONCLUSIONS: This indicates a delayed immunization recovery where past performance no longer reliably dictates the future due to logistical saturation. The immunization system of Mexico has reached a state of prolonged vulnerability, and there is an urgent need for structural reinvestments and catch-up campaigns to prevent the re-establishment of endemic measles transmission.