G. An1, G. An1 1University Of Chicago,Surgery,Chicago, IL, USA
Introduction: The social, political and economic factors involved in becoming a Trauma Center (TC) are complex, invariably involving varied goals, expertise and expectations across a range of stakeholders. The failure to objectively optimize across these factors can have catastrophic consequences on the operations of institutions aiming to become a TC. Operational viability must be a precondition if a hospital is to serve its community, and should represent a fundamental constraint on the planning for a TC. Traditional data-centric analyses cannot transparently generate the prospective scenarios needed to forecast the consequences of planning decisions. The generation of such scenarios requires the representation of health system population and process dynamics that: 1) allows for the modular representation of system components at varying levels of spatial-temporal granularity and 2) facilitates transparency by incorporating stakeholder involvement and interaction. Agent-based modeling has been extensively utilized to aid in decision analysis of multi-component/actor systems in business and social systems. Presented herein is an agent-based modeling framework for hospital operations that can be potentially expanded to the specifics of implementing a trauma center within an existing institution.
Methods: An abstracted Hospital Agent-based Model (HABM) was spatially sectioned into the emergency department (ED), radiology, OR, ICU and general care units. Individual patients and healthcare providers are represented as individual computational agents located in and moving among the regions of the hospital. Patient acuity was represented by a weighted stochastic likelihood of adverse events affected by provider response. Economic costs and returns were assigned to the actions of the various agents. Simulation experiments were performed with differing trauma populations and resourcing plans to identify process bottlenecks and critical operational tipping points between viable and non-viable scenarios.
Results: The HABM generated spatio-temporal dynamics that could account for diurnal, weekly and seasonal variation in patient type and volume. Simulation outputs of patient and economic outcomes visualized decision trade offs and the impact on non-trauma care. Robust process bottlenecks were identified in: ED patient flow, non-emergent OR requirements and surgical subspecialty resources.
Conclusion: The delivery of Trauma care is a complex, multi-factorial process that has cascading effects on hospital operations. The HABM can dynamically represent a wide range of processes and data types currently utilized in hospital operations research, and serve as a participatory, interactive platform for “virtual Kaizen” scenario exploration among stakeholders. The transparency of the underlying assumptions and expectations provided by the HABM may also serve to aid in community, policy and political engagement.