Forecasting of Acute Upper Respiratory Tract Infections based on LSTM Model

Acute upper respiratory tract infections (URTI) are among the most common health challenges globally, and the need for accurate forecasting has become increasingly important, especially in regions facing healthcare disruptions. This study focuses on the Kharkiv Oblast in Ukraine, a region significantly affected by the ongoing Russian full-scale invasion. The aim is to develop and implement a Long Short-Term Memory (LSTM) model for predicting URTI outbreaks. Monthly URTI incidence data from January 2013 to April 2024, provided by the Kharkiv Oblast Centre for Disease Control and Prevention, were used for training and validating the model. The LSTM model was chosen for its ability to capture long-term dependencies in sequential data, a crucial factor for disease forecasting. The model achieved high accuracy, with a Mean Squared Error (MSE) of 0.00271 and Mean Absolute Percentage Error (MAPE) of 1.83183 for scaled data and 19.02% for inverted data. The results demonstrate the model’s effectiveness in forecasting seasonal URTI trends and provide a critical tool for public health planning in conflict-affected regions. The study concludes that LSTM models offer valuable insights for disease forecasting and resource allocation, and future research should focus on integrating additional data sources and improving model generalizability to other regions.