Data-driven optimization for nurse scheduling and rescheduling problem

Abstract

Nurse scheduling in hospitals is a highly constrained and uncertain task. While traditional optimization models can generate feasible baseline schedules, they often fail to account for unplanned disruptions such as last-minute absences. These absences compromise care quality, create workload imbalances, and force costly last-minute adjustments. Existing models rarely integrate predictive insights or proactive mechanisms to handle such volatility. The core challenge addressed in this work is to design a scheduling and rescheduling system that anticipates and reacts to daily absences with minimal disruption, while maintaining fairness, regulatory compliance, and staffing quality.