StayLTC offers a cost-effective multimodal framework for predicting how long patients will stay in hospitals, using advanced continuous-time neural networks.

• Combines structured EHR data with clinical notes for comprehensive patient assessment
• Leverages Liquid Time-Constant Networks for improved temporal modeling of patient trajectories
• Demonstrates superior performance compared to traditional prediction methods
• Enables real-time forecasting capabilities essential for hospital resource management

This innovation matters because accurate LOS prediction helps healthcare facilities optimize resource allocation, reduce costs, and improve patient care planning. Hospital administrators can make more informed decisions about staffing, bed management, and financial planning.

StayLTC: A Cost-Effective Multimodal Framework for Hospital Length of Stay Forecasting