Predictive Maintenance Scheduling for Critical Medical Equipment Fleets to Minimize Downtime and Lifecycle Costs

Abstract

This paper addresses the design and implementation of a predictive maintenance scheduling framework for critical medical equipment fleets. The methodology integrates stochastic degradation modeling and fleet‐level optimization under resource and reliability constraints. We employ a compound stochastic process combining gamma degradation increments and a non‐homogeneous Poisson shock model to characterize equipment health trajectories. Residual failure time distributions are estimated via Bayesian updating and used to derive optimal maintenance decision thresholds. A mixed‐integer nonlinear programming (MINLP) formulation is developed to minimize expected downtime, maintenance expenditures, and patient‐impact penalties over a multi‐period horizon. The objective function accounts for labor costs, spare parts inventory holding, and risk‐adjusted service level requirements. To solve the high‐dimensional problem, we propose a rolling horizon approach coupled with Lagrangian relaxation and Benders decomposition, enabling efficient handling of coupling constraints on technician capacity, spare part lead times, and clinical blackout windows. Extensive simulation experiments on a representative hospital equipment fleet demonstrate a 35\% reduction in unscheduled downtime and a 20\% decrease in lifecycle costs compared to time‐based scheduling. The methodology supports real‐time integration with hospital information systems to dynamically update schedules as clinical demands evolve. Scalability analysis and sensitivity studies confirm the robustness of the framework under varying sensor noise levels, cost parameter uncertainty, and network topologies. The results highlight the potential of combining prognostics with advanced optimization for high‐stakes healthcare asset management.