Machine Learning–based Prediction of Postoperative Pancreatic Fistula Following Pancreaticoduodenectomy

Abstract

Objective: To develop a novel machine learning (ML) model to predict clinically relevant postoperative pancreatic fistula (CR-POPF) following pancreaticoduodenectomy (PD). Summary Background Data: Accurate prognostication of CR-POPF may allow for risk stratification and adaptive treatment strategies for potential PD candidates. However, antecedent models, such as the modified Fistula Risk Score (mFRS), are limited by poor discrimination and calibration. Methods: All records entailing PD within the 2014-2018 ACS NSQIP were identified. Additionally, patients undergoing PD at our institution between 2013 and 2021 were queried from our local data repository. An eXtreme Gradient Boosting (XGBoost) model was developed to estimate the risk of CR-POPF using data from the ACS NSQIP and evaluated using institutional data. Model discrimination was estimated using the area under the receiver operating characteristic (AUROC) and precision recall curve (AUPRC). Results: Overall, 12,281 and 445 patients undergoing PD were identified within the 2014-2018 ACS NSQIP and our institutional registry, respectively. Application of the XGBoost and mFRS scores to the internal validation dataset revealed that the former model had significantly greater AUROC (0.72 vs. 0.68, P<0.001) and AUPRC (0.22 vs. 0.18, P<0.001). Within the external validation dataset, the XGBoost model remained superior to the mFRS with an AUROC of 0.79 (95% CI 0.74-0.84) versus 0.75 (95% CI 0.70-0.80, P<0.001). In addition, AUPRC was higher for the XGBoost model, compared to the mFRS. Conclusions: Our novel ML model consistently outperformed the previously validated mFRS within internal and external validation cohorts, thereby demonstrating its generalizability and utility for enhancing prediction of CR-POPF.

Preoperative identification of early extrahepatic recurrence after hepatectomy for colorectal liver metastases: A machine learning approach

“…33,56 For example, XGBoost-based models have been used to predict risk of cardiac surgery-associated acute kidney damage (CSA-AKI) and postoperative pancreatic fistula with high discriminatory accuracy. 27,57 In the current study, the proposed XGBoost-based predictive model outperformed the conventional model reported by Wensink et al with respect to both c-index and time-dependent AUC, even though both models included common variables such as primary tumor location, T and N categories, tumor size, and tumor number (Supplemental Figure S3). 6 The results of the current study should be interpreted in light of several limitations.…”

“…26 XGBoost, an advanced gradient-boosting framework, builds an ensemble of decision trees to enhance predictive accuracy. 27 Candidate variables were selected based on the previous literature, including primary tumor location, T and WORLD JOURNAL OF SURGERY -2761 N categories, KRAS status, DFI, CEA levels, administration of NAC, and TBS-all of which can be discerned in the preoperative period. 6,17,[28][29][30][31] To reduce overfitting, a 10-fold cross-validation was employed on the entire cohort.…”

“…32 The improved performance of the XGBoost model likely derived from its decision tree architecture, which had the ability to independently capture nuanced relationships between variables. 27 Moreover, the XGBoost statistical approach may be superior to other ML techniques in terms of prediction accuracy. 33,56 For example, XGBoost-based models have been used to predict risk of cardiac surgery-associated acute kidney damage (CSA-AKI) and postoperative pancreatic fistula with high discriminatory accuracy.…”

Predictive model for very early recurrence of patients with perihilar cholangiocarcinoma: a machine learning approach

“…The XGBoost-algorithm-powered survival model was employed to develop a predictive model of VER following curative-intent resection of pCCA ( 27 ). XGBoost is an advanced implementation of the gradient boosting framework designed for speed and performance ( 28 ). This approach builds an ensemble of decision trees in which each new tree corrects the errors of the previous trees, resulting in a robust predictive model ( 28 ).…”

A nomogram for constructing a multidimensional postoperative pancreatic fistula for pancreatic tumors:Based on Systemic Inflammatory Biomarkers

“…guidance of clinicians, better preoperative intervention for patients, and improve the prognosis of patients [70] [71] [72].…”