Machine Learning-based Model for Predicting Postoperative Complications among Patients with Colonic Perforation: A...

Area of Science:

• Colorectal Surgery
• Surgical Outcomes
• Machine Learning in Medicine

Background:

• Colonic perforation surgery is associated with high morbidity and mortality.
• Predictive models for postoperative complications are lacking.
• Improving preoperative risk stratification is crucial for patient outcomes.

Purpose of the Study:

• Identify risk factors for complications after colonic perforation surgery.
• Develop a machine learning-based predictive model for these complications.
• Enhance preoperative risk assessment and clinical decision-making.

Main Methods:

• Retrospective study of 51 patients undergoing emergency colorectal perforation surgery.
• Analysis of preoperative indicators including lactate and albumin levels.
• Application of Classification and Regression Trees (CART) for predictive model development.

Main Results:

• Overall complications occurred in 62.7% of patients.
• High lactate levels and hypoalbuminemia were significant predictors of complications.
• The CART model, prioritizing albumin and lactate, achieved an AUC of 0.830.

Conclusions:

• Preoperative albumin and lactate levels are valuable predictors of postoperative complications.
• The CART model provides optimal cutoff levels for these indicators.
• These findings facilitate easier clinical use for predicting complications in colonic perforation surgery.