Demand Forecast for Human Papillomavirus Vaccine Based on the Prophet-LSTM Model

The human papillomavirus (HPV) vaccination is a critical component of China's public health prevention strategy. As vaccination programs expand, fluctuating demand poses significant challenges to inventory management: excessive stock ties up substantial capital, while insufficient supply may disrupt vaccination services. Given the high cost of HPV vaccines, accurate demand forecasting is essential for optimizing inventory, reducing waste, and improving vaccine accessibility. Traditional time-series methods alone are insufficient to address this complex issue, prompting this study to propose a data-driven hybrid forecasting approach for enhanced prediction accuracy. This study developed and compared five forecasting models: SARIMA, LSTM, Prophet, SARIMA-LSTM hybrid, and Prophet-LSTM hybrid. Experimental results demonstrate that the Prophet-LSTM model effectively captures the temporal patterns of HPV vaccine demand, achieving superior prediction performance with an R2 of 0.7336, RMSE of 6.6571, MAE of 4.7616, and MAPE of 40.95% on the test set. Furthermore, inventory cost simulations reveal that the Prophet-LSTM model significantly reduces inventory costs, providing vaccination centers with a scientific basis for optimizing stock management. The findings offer a robust demand forecasting tool for vaccination centers, supporting optimized inventory management and enhanced service stability, with significant implications for vaccine supply chain management.