Machine Learning Algorithms for Predicting Surgical Outcomes after Colorectal Surgery: A Systematic Review

Area of Science:

• Surgical Oncology
• Artificial Intelligence in Medicine
• Data Science in Healthcare

Background:

• Machine learning (ML) applications in healthcare are expanding, yet its role in colorectal surgery outcomes prediction remains under-explored.
• This systematic review provides a comprehensive overview of ML models designed to predict surgical outcomes in colorectal surgery patients.
• The study addresses a gap in the literature regarding the predictive capabilities of ML in this surgical specialty.

Approach:

• A systematic literature search was conducted across major databases (PubMed, EMBASE, Cochrane, Web of Science).
• Studies were included if they utilized ML models for predicting outcomes in patients undergoing colorectal surgery.
• The Probast risk of bias tool assessed the methodological quality of included ML models.

Key Points:

• 1821 studies were screened, with 31 articles included in the final analysis.
• ML algorithms, particularly radiomics, were frequently used to predict disease course and treatment response, achieving up to 91% accuracy.
• A significant limitation identified was the predominantly retrospective study design, lacking external validation and calibration.

Conclusions:

• ML models demonstrate considerable potential for predicting surgical outcomes in colorectal surgery.
• Bridging the gap between model calibration and external validation requires large-scale datasets.
• Clinical implementation studies are essential to confirm the practical utility of ML in daily colorectal surgery practice.