Deep Learning-Based Demand Forecasting for Blood Supply ‎Systems: A Case Study Using LSTM Networks ‎

Accurate demand forecasting plays a critical role in blood supply chain management due to the ‎perishable nature of blood products and uncertainty in demand patterns. This study develops a ‎Long Short-Term Memory (LSTM) neural network model to forecast blood demand using real-‎world data collected from healthcare centers in Fars province. After performing data ‎preprocessing, normalization, and time-series structuring, the LSTM model was trained and ‎evaluated using RMSE, and MAPE performance indicators. Results demonstrate that the ‎proposed deep learning approach effectively captures nonlinear temporal patterns and seasonal ‎fluctuations. Furthermore, a scenario-based framework is constructed to incorporate uncertainty ‎into forecasting outputs. The findings confirm that deep learning-based forecasting significantly ‎enhances decision-making reliability in blood supply planning.‎

کلیدواژه ها:

Blood Demand Forecasting ، LSTM ، Deep Learning ، Healthcare Supply Chain ، Time Series ‎Prediction