Objective To construct a spatiotemporal clustering and epidemic intensity threshold model for hand, foot and mouth disease (HFMD) based on surveillance data from 2014 to 2024 in Kunming, Yunnan, quantify its spatiotemporal distribution characteristics and epidemic intensity of HFMD, evaluate its applicability in disease surveillance and early warning, and provide quantitative evidence for the local prevention and control of HFMD.

Methods The weekly reported incidence data of HFMD from week 1 of 2014 to week 52 of 2024 in Kunming were collected. Spatiotemporal scan statistic was used to identify the spatiotemporal clustering patterns of HFMD from 2014 to 2024. The data during 2014–2023 were used as the historical baseline. Moving epidemic method (MEM) was used to identify epidemic periods and intensity. Cross-validation was conducted to evaluate model performance in terms of sensitivity, specificity, and Youden index, and optimal parameters were selected for model construction

Results From 2014 to 2024, the incidence of HFMD showed marked annual fluctuations in Kunming, with spring and summer as the main epidemic seasons. Six significant spatiotemporal clustering areas were detected (P<0.001). The optimal MEM parameterδ was 2.00. After cross-validation, the sensitivity, specificity, and Youden index were 0.81, 0.90, and 0.71, respectively. From 2014 to 2023, moderate epidemic intensity was found in most years. In 2024, the epidemic threshold was 595 cases, the epidemic began to start at week 24 and peaked at week 27 (988 cases), indicating a moderate epidemic level. The model predictions were highly consistent with observed trends.

Conclusions The incidence of HFMD exhibited significant spatiotemporal clustering in Kunming. The MEM model effectively identified epidemic period and analyze epidemic intensity, demonstrating good stability and early warning performance. The application of the MEM model can support the HFMD surveillance, early warning, and public health policy decision in Kunming.