Prediction model of laparoendoscopic single-site surgery in gynecology using machine learning algorithm
Jun Ma1, Jiani Yang1, Shanshan Cheng1
1Department of Obstetrics and Gynecology, Renji Hospital, School of Medicine, Shanghai Jiaotong University, Shanghai, China.
Summary
Laparoendoscopic single-site surgery (LESS) in gynecology reduces postoperative pain and scarring compared to conventional laparoscopic surgery (CLS). A machine learning risk model aids in preoperative patient selection for LESS.
Area of Science:
- Minimally invasive gynecologic surgery
- Surgical risk prediction modeling
Background:
- Minimally invasive surgery is prevalent in gynecology.
- Laparoendoscopic single-site surgery (LESS) offers potential benefits.
- A risk prediction model for LESS can guide preoperative decisions.
Purpose of the Study:
- Compare clinical outcomes of LESS versus conventional laparoscopic surgery (CLS).
- Develop and validate a LESS risk prediction model for clinical use.
- Assess patient suitability for LESS procedures.
Main Methods:
- Retrospective analysis of 1019 LESS and 1055 CLS patients.
- Comparison of key clinical indicators and outcomes.
- Evaluation of multiple machine learning algorithms for risk prediction.
Main Results:
- LESS demonstrated superior outcomes in postoperative pain (VAS) and scar scores.
- Both LESS and CLS showed advantages in specific clinical indicators (e.g., operative time, blood loss, hospital stay).
- The decision tree model achieved the highest predictive accuracy (AUC 0.77).
Conclusions:
- LESS is associated with reduced postoperative pain and improved scar cosmesis.
- A machine learning-based risk prediction model accurately identifies patients for LESS.
- The model supports individualized preoperative evaluation and enhances surgical safety in LESS.
