Improving patient flow in a hospital through dynamic allocation of cardiac diagnostic testing time slots

摘要

A cardiac diagnostic testing center (CDTC) makes real-time scheduling decisions that impact the use of its resources and the availability of telemetry-equipped beds within a hospital. Both inpatients and outpatients are frequent users of CDTC resources, and physicians prescribe one of several single-phase or multiple-phase test protocols. This complex online decision-making environment is modeled as a finite-horizon, discrete-time Markov decision process (MDP), but the growth of the state space motivates the introduction of a fast heuristic for real-time decision support. We therefore introduce a dynamic network scheduling tool which is both more flexible and more robust, making it applicable to the various configurations that may be found in practically any CDTC. We evaluate this new method computationally using simulation, comparing it to both an MDP model for small instances, and to the existing operational practice at our partner hospital for more realistic sized problems.