3139 - Quantifying Clinical Intuition in oART: Survey-Derived Development and Retrospective Validation of a Web-Based Adaptive Benefit Framework for Online Adaptive Radiotherapy

Purpose/Objective(s): Online adaptive radiotherapy (oART) improves target coverage and organ-at-risk sparing but remains resource-intensive, operationally complex, and patient selection is largely subjective and institution-dependent. We developed and validated a quantitative framework to estimate perceived clinical utility of oART by translating expert judgment into a reproducible decision-support tool.

Materials/Methods: A multi-institutional survey of clinicians experienced in oART identified and weighted indications for oART. The framework was based on Multi-Criteria Decision Analysis and Multi-Attribute Utility Theory, creating a structure for converting surveyed consensus into a quantitative model. While benefit assessment reflects expert perception, the weighting methodology provides an objective and reproducible aggregation process. Respondents distributed 100 points across nine consolidated indication categories and rated individual indications using a 5-point Likert scale. Indication importance was calculated as the normalized mean rating multiplied by category weight to derive global weights. These were incorporated into a linear additive model to generate a composite Adaptive Benefit Score (ABS), categorized as low (0.00–0.29), moderate (0.30–0.59), or high (0.60–1.00). Agreement among respondents was quantified using an intraclass correlation coefficient (ICC), measuring consistency of weighting across experts. The model was implemented as a web-based decision-support tool that generates ABS estimating anticipated benefit of oART. ABS was retrospectively tested in 20 patients: 10 with consensus-determined adaptive benefit and 10 transferred off oART due to limited perceived benefit.

Results: 30 respondents from 14 institutions completed the survey. Category weighting demonstrated high inter-expert reliability (ICC = 0.91), indicating strong consistency in relative prioritization across institutions. Highest-weighted categories included Clinical Trial/Patient-Centered Considerations, Tumor and Site Variability, and Critical Structure Proximity/Re-irradiation. Retrospectively, ABS differentiated patients based on consensus-determined perceived oART benefit. Mean ABS was higher in patients perceived to benefit from oART compared with those with minimal perceived benefit (0.78±0.10 vs 0.48±0.20), with high scores concentrated in the benefit cohort and low-to-moderate scores in those transferred off workflow.

Conclusion: This study demonstrates the feasibility of converting subjective expert perceptions of adaptive utility into a transparent, quantitative, and clinically deployable framework. Although benefit reflects expert judgment rather than objective outcome metrics, the structured weighting approach enables reproducibility and standardization across institutions. The ABS platform provides a scalable foundation for consistent oART selection and future validation against objective endpoints.