INTERNATIONAL JOURNAL OF ENGINEERING MANAGEMENT AND
SOCIAL SCIENCES

Queuing Theory as the Interpretable Core of Adaptive Hospital Operations: A Theory-Building Synthesis and the LASI Framework

Hospital patient flow and resource allocation are problems of decision-making under stochastic, non-stationary demand in a complex adaptive system, yet analytical queuing models, simulation, and artificial intelligence (AI) are treated as competing techniques, and no theory specifies how an interpretable core could safely constrain an adaptive one. This paper makes a theory-building contribution. It formalizes hospital operations as a single controlled-stochastic-process problem — defending that abstraction against network-optimization, partially-observable, cyber-physical, and decentralized alternatives — of which the analytical, simulation, and learning paradigms are three approximation strategies. From four functional requirements it derives the LASI framework (Layered Analytical–Simulation–Intelligence), argues each layer’s indispensability and the necessity of their dependency ordering, and specifies the core mechanism as a constrained-Markov-decision-process safety filter. Its most developed contribution is the validity-region mechanism: the analytical core may bound learning only where the core is validated, with the boundary estimated continuously by simulation through an analytical–reality divergence measure, and authority transferring between the analytical and simulation layers by a hysteresis rule. Governance is elevated from a passive gate to an active supervisory controller that resolves inter-layer conflicts and sets a responsible decision weight — the delegable authority to learning that keeps violation risk below a governance tolerance. Seven falsifiable propositions with boundary conditions state the theory. The synthesis repositions queuing theory not as a stage superseded by AI but as the interpretable, auditable core of accountable adaptive hospital operations.