Predictive accuracy of computer-generated padua nephrometry scores based on continuous variables compared with categorical computer-generated scores and human-generated scores in predicting oncologic and perioperative outcomes.

• DOI: 10.1200/jco.2023.41.6_suppl.624
• Publication Date: 2023-03-14T15:22:32Z
• Authors: Tarik Benidir; Andrew Wood; Nour Abdallah; Erick M. Remer; Steven Campbell; Christopher J. We...
• Keywords: Categorical variable; Continuous variable

624 Background: The Preoperative Aspects and Dimensions Used for Anatomical Classification (PADUA) score is a validated predictor of patient’s perioperative outcomes following partial nephrectomy. use fully automated, unbiased time sensitive PADUA scoring system novel concept that may help circumvent PADUA’s widespread adoptability. We sought to automate the preoperative computed tomography scans (CT) using machine-learning-generated (P-AI). In doing so, we aimed compare P-AI’s ability predict meaningful oncologic as compared human generated scores both categorical (P-H) continuous variables (P-AI+). Methods: 300 patients with pre-operative CT were identified from cohort 544 consecutive undergoing surgical extirpation suspected renal cancer at single institution. A deep neural network approach was used automatically segment kidneys tumors geometric algorithms developed estimate each component based on segmented regions Tumors also manually scored by medical personnel blinded P-AI (P-H). P-AI+ P-H logistic regression receiver operating characteristics (ROC) curve analyses areas under (AUC). Results: Median age 60 years, 40% female. tumor size 4.2 cm, 91.3% had malignant tumors, including 27% 37% high-stage high-grade, respectively. Both able need nephrectomy (p < 0.001). From an standpoint, presence malignancy, high grade stage disease 0.004) although ROC curves not different one another). Interestingly, when left rather than ordinal (P-AI+), this automated type (AUC 0.88), malignancy 0.67), indolent 0.79) 0.77)/high 0.82), better P-H. Conclusions: When viewed along spectrum, AI (P-AI+) yields predictive superior scores. If nephrometry can be calculated automatically, there no longer simplify equations system, provides reliable in manner expert evaluation. Further prospective work reproducibility other centers encouraged.