Risk-based clinical scheduling tool for congenital cardiac catheterization procedures

In centers with multiple catheterization laboratories and other complex procedural units, cases are scheduled to occur simultaneously, resulting in shared resources and compounding risk factors within the care environment. Additional complexity arises from the heterogeneous nature of procedure types, the frequency of cases and turnover, and the resource requirements necessary to care for these patients. This complexity necessitates an innovative approach to scheduling that enhances both safety and efficiency. In collaboration, The MITRE Corporation and the cardiac catheterization laboratory at Boston Children's Hospital (BCH) developed a tool that allows decision-making about scheduling cases to be based on risk and resource utilization. The aim of this study is to leverage to validate human-interpretable scheduling heuristics that decrease system-level risk, increase system-level efficiency, and are easily integrated into existing scheduler workflows. The Points Split heuristic produced schedules with much fewer unbalanced days compared to the Baseline and Points heuristics. The median count of unbalanced days per year for the Points Split heuristic was 7, compared to 78 and 110 unbalanced days per year for the Points and Baseline heuristics, respectively. This machine learning-enhanced scheduling tool effectively aligns patient risk with resource availability, thereby enhancing operational efficiency and safety in pediatric cardiac catheterization labs. The rule-based scheduling heuristic was also found to be robust to a variety of lab configurations, case arrival rates, and patient population conditions. The approach holds promise for broader application in complex medical environments where procedure scheduling impacts patient outcomes. • Developed a machine learning-enhanced scheduling tool for congenital cardiac catheterization that integrates predictive modeling and simulation-based heuristics to align procedural risk with resource availability, improving both safety and operational efficiency. • Introduced the “Points Split” heuristic, a novel rule-based scheduling approach that separates case risk and duration constraints, significantly reducing unbalanced scheduling days per year compared to current practice. • Applied discrete-event simulation to evaluate scheduling policies, capturing real-world complexities such as case start delays, adverse events, ICU admissions, and provider workload, offering a robust framework for procedural planning. • Leveraged multi-source datasets, including C3PO and institutional scheduling data, to generate synthetic case populations, enabling generalizable analysis of system performance under varying case complexity and arrival rates. • Demonstrated robust performance under high volume and complexity with reduced ICU risk and lab overtime.