Spatial and temporal analysis of dengue incidence in Northern Thailand: a 13-year retrospective study (2012–2024)

Abstract Background Dengue fever remains a major public health concern in the northern region of Thailand, where periodic outbreaks impose substantial health and economic burdens. Understanding spatial and temporal patterns of dengue incidence rate is essential for improving surveillance, forecasting, and targeted control strategies. Objectives This study examined the spatial and temporal dynamics of dengue incidence rate across eight provinces in the northern region of Thailand and developed predictive models to support early warning and prevention efforts. Methods Monthly dengue case data from 2012 to 2024 were obtained from the national surveillance system and aggregated at the provincial level. Spatial analyses included the calculation of incidence rates, spatial empirical Bayesian (SEB) smoothing, and assessment of spatial clustering using Moran’s I and local indicators of spatial association (LISA). Temporal analyses employed seasonal-trend decomposition and seasonal autoregressive integrated moving average (SARIMA) modeling to forecast dengue incidence rate at regional and provincial levels. To ensure normal transmission patterns, incidence rates during the COVID-19 period were excluded. Model performance was evaluated using mean absolute error (MAE), root mean square error (RMSE) and mean absolute percentage error (MAPE). Results Dengue incidence rate showed clear seasonality, with annual peaks during the rainy season (June–August). Spatial heterogeneity was evident, with Chiang Mai and Mae Hong Son consistently exhibiting high incidence rate and significant spatial clustering. The SARIMA (2,0,0)(2,1,0) 12 model accurately captured regional dengue dynamics, yielding MAE, RMSE, and MAPE of 2.93, 3.30, and 31.86%, respectively. Forecasts for 2025 indicated a regional peak in July (88.07 per 100,000), with the highest provincial peaks expected in Lamphun (74.76, August) and Chiang Mai (60.25, July). Conclusions Integrating spatial clustering with temporal forecasting enhances understanding of dengue transmission dynamics and supports locally tailored, data-driven interventions. These findings provide actionable insights for strengthening surveillance, optimizing resource allocation, and improving vector control strategies in endemic regions.