Practice Makes Faster In Ctgart: Objective Evidence That Experience Improves Adaptor Efficiency
Abstract
Purpose
CT-guided adaptive radiotherapy (CTgART) introduces substantial workflow complexity and resource demands. While clinical experience is expected to improve efficiency, the magnitude and structure of adaptive learning have not been well quantified using objective data. This study quantifies programmatic and individual learning effects, evaluates complexity-dependent efficiency gains, and characterizes steady-state performance in a comprehensive CTgART program.
Methods
Workflow data was retrospectively extracted from Varian Ethos log files for adaptive fractions delivered between 8/2021-12/2025. Extracted metrics included system-reported time points corresponding the adaptive workflow. Treatment complexity was classified using a data-driven model of intent-specific median door-to-door(D2D) time categorizing fractions as Simple, Moderate, or Complex. Learning was evaluated using a stratified experience-vs-performance trend and a time to steady-state analysis. Steady state was defined as the earliest fraction an adaptor’s median D2D time remained within±5% of their late-phase median for 20 consecutive fractions.
Results
A total of 4,252 adaptive fractions from 286 patients were analyzed. Overall median D2D time was 33.8(13.4–123.5)min. Median D2D, increased with treatment complexity, 29.5,34.5,and 48.3 min for Simple, Moderate, and Complex sessions, respectively. From 2022-2025, annual adaptive fraction volume increased for Moderate(267-366) and Complex(35-239) treatments, while Simple cases decreased(769-641). Over the same period, median D2D treatment time decreased for Moderate(39.2 to 34.5 min) and Complex(55.6 to 48.3 min). Across individual adaptors, experience-stratified comparisons demonstrated a median D2D improvement of 6.9%(−4.6 min). Experience vs performance trend analysis demonstrated a median improvement rate of 0.20 min per ln(Session). Moderate treatments demonstrated the largest efficiency gains(−11.9%), while Simple had minimal change(−1.7%). Steady-state analysis demonstrated that adaptors reached stable performance after a median of 30 fractions.
Conclusion
This study demonstrates that for CTgART, efficiency improves with experience for both the program and individual. Learning effects are strongly dependent on treatment complexity and Experience. Quantitative benchmarks can inform training expectations, staffing models, and operational planning in CTgART programs.