Forecasting Global Tuberculosis Incidence Rates Using a Multi‐Variate Input, Multi‐Region, Multi‐Step Output Integrated Model: SP‐CTLA

ABSTRACT Tuberculosis (TB) remains one of the leading global infectious diseases, causing millions of deaths each year, despite extensive efforts to control its spread. It continues to pose a significant public health challenge, particularly in low‐ and middle‐income countries, where factors such as poverty, malnutrition, and inadequate healthcare systems exacerbate the situation. As a result, precise and timely forecasting of global TB incidence is crucial to achieving the World Health Organization's (WHO) TB End Strategy and the United Nations' (UN) Sustainable Development Goals (SDGs). However, existing studies primarily focus on predicting TB incidence within a single region, often neglecting the broader regional or cross‐border factors that can significantly impact TB trends. This limitation arises because these studies fail to consider the spatial dependencies between countries, where the TB incidence rates of neighboring countries can influence each other. To address this issue, this study introduces an innovative approach by integrating spatial information, including national coordinates and the TB incidence rates of neighboring countries, along with socioeconomic, demographic, and environmental factors. Based on these factors, a multi‐variable input, multi‐country output, and multi‐time‐step forecasting model—Spatial‐CNN‐Transformer‐LSTM‐Attention (SP‐CTLA)—has been developed. The model demonstrated robust performance in both testing and Monte Carlo cross‐validation, with the following results: RMSE of 16.469, MAE of 9.282, and MAPE of 21.560%. In the testing set, the results were RMSE of 15.439, MAE of 8.327, and MAPE of 17.747%. The model forecasts TB incidence rates for 163 countries from 2024 to 2026, generating spatial distribution maps and identifying the top ten countries with the most significant increases and decreases in incidence. The projections suggest that 25 countries will achieve the WHO's 2025 TB End Strategy ahead of schedule, by 2026. This study identifies the top three factors influencing TB incidence rates: the incidence rates of neighboring countries, mean surface temperature, and the Mean NDVI value. The findings offer a novel forecasting model for global TB incidence rates, providing valuable insights for effective TB control strategies and contributing to the achievement of the WHO's TB End Strategy and the UN's SDGs.