Authors: Sahil Arun Sahane, Suhani Sharma, Amay Prasad Sabnis, Suhani Singh, Assistant Professor Rahul B. Mandlik
Abstract: Efficient management of critical hospital resources such as intensive care unit (ICU) beds, oxygen supply, and medical staff has become a major challenge, particularly during large-scale healthcare emergencies. Conventional hospital management systems are largely reactive and often fail to anticipate sudden surges in patient demand, resulting in delayed responses and resource shortages. This paper presents MediCast, an AI-driven hospital resource forecasting and decision support system designed to predict ICU bed occupancy and oxygen demand in advance while supporting optimized resource allocation. The proposed framework employs Long Short-Term Memory (LSTM) networks for time-series forecasting of ICU admissions and oxygen consumption trends, and XGBoost models for learning complex patterns from structured hospital data. Based on the predicted demand, an optimization layer assists in efficient allocation of beds and staff resources to reduce overload and improve preparedness. The system also provides an interactive dashboard for real-time visualization of predictions, alerts, and analytical insights, enabling hospital administrators to take proactive decisions. By integrating predictive analytics and optimization within a unified platform, MediCast enhances operational efficiency, minimizes critical resource shortages, and supports data-driven healthcare management in high-demand scenarios.
