Article Text
Abstract
Introduction/Background Surgical cytoreduction for epithelial ovarian cancer (EOC) is a complex procedure, whereas that intra-operative surgical decision-making remains a core feature. Explainability Artificial Intelligence (XAI) could potentially interpret the influence of human factors on the surgical effort for the cytoreductive outcome in question.
Methodology The retrospective cohort study evaluated 560 consecutive EOC patients who underwent cytoreductive surgery between January 2014 and December 2019 in a single UK institution. The eXtreme Gradient Boosting (XGBoost) was employed to develop the predictive model including patient- and operation-specific features, readily available in tertiary centers, and novel features reflecting human factors in surgical heuristics. The area under the curve (AUC) was used to evaluate model performance. The SHapley Additive exPlanations (SHAP) framework was used to provide global and local explainability of the predictive model.
Results A surgical complexity score (SCS) cut-off value of five was calculated using a receiver operator characteristic (ROC) curve, above which the probability of incomplete cytoreduction was more likely (area under the curve [AUC]=0.644; 95% [CI]=0.598–0.69; sensitivity and specificity 34.1%, 86.5%, respectively; p=0.000). The XGBoost model performance for the prediction of the above threshold surgical effort outcome was satisfactory (AUC=0.77; 95%[CI] 0.69–0.85; p<0.05). ‘Turning points’ showing preference towards above-given threshold surgical effort included; consultant surgeons with <12 years of experience, age <53 years old, who, when attempting primary cytoreductive surgery, recorded the presence of ascites, an Intraoperative Mapping of Ovarian Cancer score >4, and a Peritoneal Carcinomatosis Index >7, in a surgical environment with optimization of infrastructural support.
Conclusion Surgical intra-operative decision-making is critically layered upon situational awareness and the impact of human factors. We demonstrated a fine balance between predictive accuracy and descriptive interpretability. Using XAI, we provided a two-layered explainability and we pinpointed the most salient feature interactions. Selected decreased surgical effort may be associated with surgeon age.