A probabilistic framework for risk-bounded patient-specific quality assurance in volumetric modulated arc therapy based on measured and calculated gamma pass rates

This study introduces a probabilistic framework for patient-specific quality assurance (PSQA) in volumetric modulated arc therapy (VMAT), using gamma pass rates obtained from both measurement-based and independent calculation-based PSQA. The model quantifies the probability that treatment plans classified as acceptable by an independent dose calculation (IDC) algorithm would nevertheless fail measurement-based verification. This probability, referred to as the false omission rate, represents the residual risk of skipping measurements for plans classified as unproblematic by the calculation. Receiver operating characteristic (ROC) analysis demonstrated limited discriminative performance of the IDC, underscoring that reliance on calculation alone entails non-negligible miss probabilities. In a retrospective cohort of 1346 clinical VMAT plans across multiple anatomical sites, false omission rates ranged from below 1% in well-represented groups such as prostate to above 20% for sparsely represented treatment regions. Based on these empirically derived miss probabilities, the framework allows the definition of measurement intervals that constrain the cumulative probability of undetected plan failure below a predefined institutional threshold. In this way, PSQA strategies involving reduced measurement schedules for selected plans can be quantitatively assessed and constrained within explicit probabilistic safety limits. The proposed method is transparent, independent of complex machine learning models, and directly applicable to empirical QA datasets. By making the risk of undetected failure explicit and quantifiable, it enables structured, risk-informed PSQA policies grounded in defined safety constraints.