Machine learning applications in upper gastrointestinal cancer surgery: a systematic review
Mustafa Bektaş1, George L Burchell2, H Jaap Bonjer3
1Surgery, Amsterdam UMC Location Vrije Universiteit Amsterdam, De Boelelaan 1117, Amsterdam, The Netherlands. m.bektas@amsterdamumc.nl.
Surgical Endoscopy
|August 11, 2022
Summary
Machine learning aids in predicting outcomes for upper gastrointestinal cancer surgery. Further prospective studies are needed to confirm these findings for machine learning applications in surgical oncology.
Area of Science:
- Surgical Oncology
- Medical Informatics
- Artificial Intelligence in Medicine
Background:
- The application of machine learning (ML) in upper gastrointestinal surgery for malignancies is not well-documented.
- This systematic review provides a comprehensive overview of ML applications in this surgical field.
Approach:
- A systematic literature search was conducted across major databases (PubMed, EMBASE, Cochrane, Web of Science).
- Included studies focused on ML in upper gastrointestinal surgery for malignancies, with methodological quality assessed using the Cochrane risk-of-bias tool.
- Predictive performance of ML models was evaluated using accuracy and area under the curve metrics.
Key Points:
- 27 studies met inclusion criteria from 1821 articles, with most exhibiting moderate risk of bias.
- Neural networks, multiple machine learning algorithms, and random forests were the most common ML techniques used.
- ML applications focused on predicting metastasis, risk factors, survival, postoperative complications, TNM staging, chemotherapy response, resectability, and optimal therapy.
Conclusions:
- Machine learning algorithms show potential in predicting postoperative complications and disease course in upper gastrointestinal cancer surgery.
- The retrospective nature of current ML studies necessitates prospective trials to validate these predictive applications.
