Forecasting Institutional LINAC Utilization in Response to Varying Workload.

Area of Science:

• Medical Physics
• Radiotherapy Operations
• Health Services Research

Background:

• Unforeseen events like pandemics can disrupt radiotherapy service delivery and resource allocation.
• Accurate forecasting of linear accelerator (LINAC) utilization is crucial for efficient radiotherapy operations.
• Existing methods may not adequately address short-term variations in LINAC workload.

Purpose of the Study:

• To develop and validate a predictive model for short-term LINAC utilization.
• To forecast LINAC usage over a 15-working day horizon.
• To provide a tool for optimizing resource planning in radiotherapy departments.

Main Methods:

• A multiple linear regression model was developed using computed tomography (CT)-simulation data and prior week LINAC appointment booking rates.
• The model was trained and validated on institutional data to forecast LINAC utilization.
• Performance was compared against moving average and exponential smoothing techniques.

Main Results:

• The model achieved low forecasting errors: 3.3% (day 5), 5.9% (day 10), and 7.2% (day 15) on the training set.
• Accuracy in identifying significant LINAC utilization variations (≥5%) was 69% (day 5), 62% (day 10), and 60% (day 15).
• Validation dataset results showed comparable performance, demonstrating model robustness.

Conclusions:

• The developed linear regression model provides accurate short-term forecasting of LINAC utilization.
• The model effectively uses CT simulation data and booking rates as predictive inputs.
• This forecasting tool has been integrated into an institutional dashboard for practical application in resource management.