Artificial intelligence algorithms for predicting post-operative ileus after laparoscopic surgery.

Area of Science:

• Surgical Oncology
• Artificial Intelligence in Medicine
• Machine Learning

Background:

• Postoperative intestinal obstruction (POI) is a significant complication following laparoscopic colorectal cancer surgery.
• Accurate prediction of POI risk is crucial for individualized prevention and treatment strategies.

Purpose of the Study:

• To develop and validate an artificial intelligence-based predictive model for postoperative intestinal obstruction in patients undergoing laparoscopic colorectal cancer surgery.
• To identify key risk factors contributing to the development of POI.

Main Methods:

• A predictive model was constructed using machine learning and deep learning algorithms in Python.
• Subjects were divided into training (70%) and testing (30%) sets.
• Model performance was evaluated using metrics including AUC, accuracy, precision, recall, and F1 scores.

Main Results:

• Opioid use, anesthesia duration, and body weight were identified as the top three risk factors for POI.
• XGBoost (XGB) and Decision Tree algorithms demonstrated the highest accuracy (0.807) and precision (0.500).
• Gradient Boosting and XGB algorithms achieved the best AUC values (0.678 and 0.638, respectively).

Conclusions:

• XGBoost and Decision Tree are the most effective algorithms for predicting ileus risk after laparoscopic colorectal cancer surgery.
• This AI-driven approach offers novel insights and supports clinical decision-making for managing POI.
• The study highlights the potential of machine learning to improve patient outcomes in colorectal cancer care.