Team Twente's Hybrid AI Approach Takes Third in Healthcare Timetabling Competition
Team Twente's Hybrid AI Approach Takes Third in Healthcare Timetabling Competition
Team Twente, representing the University of Twente, has secured third place in the Integrated Healthcare Timetabling Competition (IHTC) 2024. Their innovative hybrid AI solution combines Mixed-Integer Linear Programming (MILP), Constraint Programming (CP), and Simulated Annealing (SA) to address the complexities of healthcare timetabling.
The competition focused on optimizing patient admissions, room allocation, nurse assignment, and operating theater planning within healthcare systems. Team Twente's approach decomposes the problem into manageable subproblems, solved iteratively and in parallel using a three-phase strategy.
The Three-Phase Decomposition Approach
Phase 1 optimizes initial patient admission and calculates lower bounds using information from other subproblems. Phase 2 uses these bounds to iteratively solve patient-day admission, patient-room assignment, patient-theater assignment, and nurse-room assignment. Phase 3 refines nurse assignment using an exact approach after a heuristic solution in Phase 2.
The solution runs in parallel, utilizing all available threads to maintain a pool of partial and complete solutions throughout the runtime. Feasible solutions to subproblems are used as inputs for subsequent subproblems, enhancing efficiency.
Implementation and Tools
The implementation, available on GitHub, is written in Python and uses the Gurobi 12 MIP solver and Google's OR-Tools for CP. The team's insights and lower bounds for benchmark instances provide valuable contributions to the field.
Researchers noted potential improvements in admission scheduling, room assignments, and nurse scheduling efficiency. Feedback mechanisms and cutting planes were suggested to enhance nurse scheduling further.
Conclusion
Team Twente's hybrid AI approach demonstrates the potential of combining optimization techniques for complex healthcare planning. Their solution offers a scalable and efficient framework for future developments in integrated healthcare timetabling.