Background : This study aims to evaluate supervised machine learning algorithms in predicting 90-day post-operative events (mortality/morbidity) in gastrointestinal and HPB surgeries.

Methods : We evaluated various supervised machine-learning classification algorithms in random forests. We used accuracy and the Receiver operating curve to compare the methods. 60% of the data were used for training, 20% for validation and 20% for testing. We defined postoperative events as any mortality and grade-3,4 clavien-dindio complications within 90 days of post-operative period. We used JASP 0.16.04 by the University of Amsterdam to run machine learning algorithms and statistical analysis.

Results : We used data from 525 patients who have undergone gastrointestinal and hepatopancreatic biliary surgery between April 2016 and December 2023. We analyzed algorithms for predicting 90 days post-operative events based on features like Major surgeries, Surgeries for malignancies, age, Intraoperative hypotension, Open vs Laparoscopic surgeries, ASA grade, Emergency surgeries, Operative time, Intraoperative blood product used, colorectal surgeries, small intestinal surgeries, HPB surgeries, and upper gastrointestinal surgeries. Test accuracies were 87.6% and Areas under the ROC curve were 0.816. Features of importance in decreasing order were Intraoperative hypotension, ASA, operative times, age, emergency surgeries, HPB surgeries, small bowel surgeries, open surgeries, colorectal surgeries and malignant surgeries.

Conclusions : Supervised machine learning algorithm random forest predicted 90 days post-operative events accurately and such models can be part of the preoperative evaluation in gastrointestinal and HPB surgeries.