EURO 2025 Leeds
Abstract Submission

2478. Healthcare Predictive Timetabling: A predictive control solver for integrated healthcare timetabling

Invited abstract in session MB-10: Integrated Healthcare Timetabling Competition I, stream Automated Timetabling.

Monday, 10:30-12:00
Room: Clarendon SR 1.06

Authors (first author is the speaker)

1. Daniele Giovanni Gioia
Deutsche Zentrum für Luft und Raumfahrt
2. Lorenzo Mazza
Dipartimento di Scienze Matematiche "Giuseppe Luigi Lagrange", Politecnico di Torino
3. Manuel MACIS
DISMA, Politecnico di Torino
4. Edoardo Fadda
Automatica e Informatica, Politecnico di Torino
5. Paolo Brandimarte
Politecnico di Torino

Abstract

Given the importance of timetabling for healthcare applications, a two-phase solution approach for the Integrated Healthcare Timetabling Problem is proposed. It addresses the Surgical Case Planning (SCP), Patient Administration Scheduling (PAS), and Nursing Resource Allocation (NRA) via Mixed-Integer Linear Programming. In the first phase, SCP and PAS are combined in a single model (SCPPAS), recognizing that the feasibility of any timetable depends on the alignment of surgical cases with patient admissions. While NRA is solved separately, its cost implications are approximated in the SCPPAS objective to ensure realistic resource considerations. The second phase solves NRA exhaustively, leveraging the previous approximation to refine staffing allocations. Since SCPPAS can become prohibitively large for real-world instances, a model predictive control strategy is adopted. Specifically, both the planning horizon and spatial dimension are restricted at each iteration, and then rolled forward over time. Smaller, more tractable subproblems are solved iteratively. This rolling approach ensures that decisions remain in a neighborhood of the feasible space, maintaining an integrated perspective across all subproblems. Preliminary computational experiments demonstrate that our method yields feasible schedules with tunable time and computation requirements.

Keywords

Status: accepted

Back to the list of papers