Accurately forecasting blood demand is crucial for maintaining optimal inventory levels in blood banks, ensuring an adequate supply for emergencies while reducing wastage. However, traditional forecasting methods often struggle to capture the intricate patterns inherent in blood demand data, leading to sub optimal inventory management. To address this, this study proposes an innovative approach to blood demand forecasting using the XGBoost machine learning model. XGBoost, a robust gradient boosting algorithm, performs well when it comes to the handling of high-dimensional data and can efficiently capture complex relationships in the data. Leveraging its capabilities, we develop a predictive model trained on historical blood demand data, considering factors such as demographic trends, seasonality, and past donation patterns. Through meticulous feature engineering and model optimization, our XGBoost-based approach aims to improve the accuracy and reliability of blood demand forecasts for blood banks. To evaluate our methodology, we conduct extensive experiments using real-world blood demand datasets from multiple blood banks. Comparative analyses against traditional methods assess the predictive performance and robustness of the XGBoost model.