Smart scheduling of arthroplasty surgery with machine learning and optimisation improves operating room utilisation.

Area of Science:

• Orthopedic Surgery
• Health Informatics
• Operations Research

Background:

• Total hip and knee arthroplasty (THA and TKA) are effective but resource-intensive procedures.
• Efficient surgical scheduling is crucial for hospital resource management and cost-effectiveness.

Purpose of the Study:

• To evaluate the efficiency of a machine learning (ML) and mathematical optimization-based surgical schedule compared to traditional methods.
• To assess the impact of ML-driven scheduling on operating room (OR) utilization and patient throughput.

Main Methods:

• Utilized data from 15,267 primary and revision THA and TKA cases (April 2012-February 2022).
• Developed procedure-specific ML models to predict operative times.
• Employed integer linear programming for OR utilization optimization and compared it with historical scheduling practices.

Main Results:

• ML models achieved a 7.1% improvement in predicting operative times compared to rolling averages.
• Optimized scheduling reduced OR underutilization by 56.2% (13.3 minutes/day) and increased case completion by 6.1% (31 OR days).
• Overtime increased by only 17.2% (3.6 minutes/day) with the ML-optimized schedule.

Conclusions:

• ML-predicted operative times and optimization strategies can significantly reduce OR idle time.
• Data-driven surgical scheduling maximizes the utilization of existing hospital resources and improves patient throughput.