目的 对黄山市结核病发病的趋势进行预测，为优化结核病防控举措提供依据。方法 从结核病管理信息系统中收集2015—2024年黄山市的结核病发病数，分别构建季节性自回归综合移动平均（seasonal autoregressive integrated moving average, SARIMA）模型、Holt-winters指数平滑模型、灰色预测GM（1, 1）模型及Prophet模型，将各模型2024年的发病预测数与实际发病数进行验证，以均方根误差（root mean square error, RMSE）、平均绝对误差（mean absolute error, MAE）、平均绝对百分比误差（mean absolute percentage error, MAPE）作为模型评价指标，选取最优模型预测2025—2026年黄山市结核病发病情况。结果 2015—2024年黄山市结核病登记发病数呈波动下降趋势，具有明显的季节性，夏季是登记报告的高峰期，冬季是低谷期；4种模型拟合效果均良好，经验证，Prophet模型预测效果最优（RMSE、MAE、MAPE值分别为5.03、4.17和10.28%）；所有模型均预测2025—2026年黄山市结核病发病数仍呈下降趋势，Prophet模型预测2025—2026年黄山市结核病登记发病数分别为462例和432例，登记发病率分别为35.08/10万和32.80/10万。结论 Prophet模型拟合效果最优，可使用该模型对黄山市结核病发病趋势进行预测。

Objective To predict the incidence trend of tuberculosis in Huangshan and to provide a basis for optimizing tuberculosis prevention and control strategies. Methods Tuberculosis incidence data from 2015 to 2024 in Huangshan were collected from the Tuberculosis Management Information System. Seasonal autoregressive integrated moving average (SARIMA) model, Holt-winters exponential smoothing model, Gray Prediction GM (1, 1) model, and Prophet model were constructed. The predicted incidence for 2024 from each model was validated against the actual reported incidence. Root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) were used as evaluation indicators to select the optimal model for predicting tuberculosis incidence from 2025 to 2026 in Huangshan. Results The registered tuberculosis incidence in Huangshan from 2015 to 2024 showed a fluctuating downward trend with distinct seasonality, peaked in summer, and reached a trough in winter. All four models demonstrated good fitting performance. Validation results revealed that the Prophet model achieved the best predictive performance (RMSE=5.03, MAE=4.17, MAPE=10.28%). All models predicted a continued decline in tuberculosis incidence in Huangshan for 2025-2026. The Prophet model predicted 462 and 432 registered tuberculosis cases in 2025 and 2026, respectively, corresponding to registered incidence rates of 35.08/100 000 and 32.80/100 000. Conclusion The Prophet model exhibits the best fitting performance and can be used to predict the trend of tuberculosis incidence in Huangshan.