Spatio-Temporal Modelling and Forecasting of the Prolonged Measles Outbreak in Romania: Insights and Challenges.

Abstract

Background/Objectives: Measles is a highly contagious viral disease that continues to have a profound effect on morbidity in Romania. Identifying temporal and spatial trends in how the disease spreads among the country's counties and regions, both in the same disease generation as well as one generation apart (2-week case lag), aided by forecasting tools, could provide valuable insights into tailoring public health interventions. Methods: A big data analysis has been performed on notified measles cases from January 2020 to December 2024 using Python v3.13 grouping cases based on location (using the Nomenclature of territorial units for statistics) and time of the onset of the disease. Results: Feedback loops among both counties and macroregions have been identified (for example Centru-Brașov and București-Ilfov with a correlation factor of 0.77) while monthly forecasting for 2025 and 2026, explored using both the SARIMA and the Holt-Winters models (MAE 1616.74 and 1281.99, respectively), shows the measles might continue to be a burden, with the Holt-Winters models exhibiting slightly more reliable monthly forecast data nationwide, helping to define a solid basis for future predictions and decisions. Conclusions: The spatial feedback loops, both interregional or within the same region, coupled with the trend of lowering vaccination rates, contribute to the persistent emergence of new measles cases which might continue throughout 2025 and 2026 based on the forecasting, distinct from previous outbreaks which followed a specific cadence.