A simulation analytics framework for system-level digital twins in hospital operations

Healthcare digital twins supporting facility planning and operation remain scarcely documented, and existing typologies lack external grounding. This paper proposes a new, theoretically grounded typology of healthcare digital twins that builds on an existing one, and documents a system-level digital twin prototype designed to size and operate a new hospital multiservice centre. Following a design science research approach, we develop two artefacts. The first is a typology crossing four hierarchical levels with three life-cycle states (design, monitoring, and operational management), defined by an adapted analytics quartet that merges descriptive and diagnostic functionalities and adds a prospective (what-if) functionality. Seven of its twelve cells contain documented twins. The second is a prototype for a multiservice centre of the Santé Québec West-Central Montreal Health and Social Services University Network, coupling RFID data capture with a hybrid agent-based/discrete-event AnyLogic model; it occupies an underexplored cell. The prototype was evaluated both in a university laboratory and on site. It generated a feasible configuration in terms of staffing, opening hours, and resource pooling, given service durations and patient volumes. Results indicate a reduction in length of stay from 266 to 75 min, i.e., 3.5 times shorter. On-site read events fed the model automatically, and the twin’s feedback rule was successfully tested. These results show that such a prototype can support operating decisions before a facility opens, while the life-cycle states of the typology frame its transition to operational management.