Robust Prediction Of Treatment Times In Concurrent Patient Care.

Area of Science:

• Healthcare Operations Research
• Clinical Informatics
• Simulation Modeling

Background:

• Outpatient centers face increasing patient volumes, necessitating efficient clinic flow.
• Shared critical resources (personnel, equipment) create interdependencies between concurrent clinics.
• Existing discrete event simulation models often lack validation for high-volume settings with shared resources.

Purpose of the Study:

• To evaluate the predictive performance of simulation models in high-volume outpatient settings with shared resources.
• To assess the impact of explicitly modeling interdependencies on model accuracy.
• To validate simulation model predictions using cross-validation techniques.

Main Methods:

• Developed a stochastic reward net model using phase-type distributions for treatment durations.
• Utilized discrete event simulation to solve the model.
• Employed cross-validation to evaluate model predictive accuracy with new data.
• Verified electronic health records (EHR) data accuracy through in-person observation.

Main Results:

• Model accuracy improved when explicitly accounting for concurrent demand on clinic staff (interdependence).
• Cross-validation demonstrated the robustness of treatment time predictions.
• The study compared two model configurations: one with concurrent demand and one with independent clinics.

Conclusions:

• Explicitly modeling interdependencies in simulation models enhances prediction accuracy for outpatient clinics with shared resources.
• Cross-validation is crucial for validating simulation models, especially when ongoing data collection is challenging.
• Accurate simulation models are vital for optimizing performance in complex, high-volume healthcare settings.