Abstract

Artificial intelligence (AI) has progressively been used in surgical decision-making, especially in emergency and trauma surgery. The Predictive Optimum Triage and Treatment Evaluator for Rapid Response (POTTER) is a novel AI-based risk assessment algorithm designed to improve preoperative planning and outcome forecasting. Nonetheless, its prediction accuracy and practical applicability remain ambiguous in comparison to established scoring systems as the American College of Surgeons Surgical Risk Calculator (ACS-SRC) and the Emergency Surgery Score (ESS). This systematic review and meta-analysis aimed to assess the predictive efficacy of POTTER in emergency general and trauma surgery, juxtaposing its accuracy with ACS-SRC in forecasting postoperative morbidity, mortality, and other critical surgical outcomes. A thorough literature review was performed to uncover papers comparing POTTER with ACS-SRC in emergency and trauma surgical contexts. A meta-analysis was conducted to evaluate discriminative capability by aggregated C-statistics and logit differences. Heterogeneity and bias risk were assessed to guarantee the robustness of the findings. The investigation revealed that although POTTER is a proficient AI-driven predictive instrument, surgeon-based evaluations (ACS-SRC and ESS) displayed enhanced discriminative efficacy in forecasting 90-day death (23% higher AUC), overall morbidity (19% higher AUC), and renal failure (21% higher AUC). Nonetheless, POTTER proved to be a valuable and comprehensible paradigm, especially in instances of septic shock and extended ventilation. Notwithstanding its promise, POTTER failed to surpass conventional surgeon-based grading systems in emergency general and trauma surgery. Additional refining and empirical validation are necessary before AI-driven models such as POTTER might supplant current risk assessment instruments.

Key words: Emergency general surgery, risk prediction, POTTER, ACS-SRC, systematic review