Artificial Intelligence in Minimally Invasive Gynecological Surgery: A Systematic Review of Task-Specific Performance...

Area of Science:

• Gynecologic Surgery
• Artificial Intelligence
• Medical Technology

Background:

• Minimally invasive gynecologic surgery (MIGS) is evolving with technological advancements.
• Artificial intelligence (AI) offers potential for improving surgical outcomes and efficiency.
• Evaluating AI's clinical readiness is crucial for its integration into surgical practice.

Purpose of the Study:

• To systematically assess the performance and clinical translational readiness of AI applications in MIGS.
• To analyze AI tools across preoperative, intraoperative, and postoperative surgical phases.
• To identify gaps in evidence and guide future AI development in gynecologic surgery.

Main Methods:

• Systematic review following PRISMA guidelines.
• Searched major databases (PubMed, Embase, etc.) for studies from 2015-2025.
• Included AI/ML models for predictive/diagnostic tasks in MIGS; excluded non-clinical/non-English studies.
• Assessed risk of bias using the PROBAST-AI framework.

Main Results:

• Eight studies (6,577 patients) evaluated AI in MIGS.
• Preoperative AI showed high sensitivity but low specificity for conditions like endometriosis.
• AI-assisted planning reduced operative time and blood loss; postoperative AI demonstrated high predictive performance for fertility outcomes.
• 75% of studies had a high risk of bias, with limited external validation.

Conclusions:

• AI applications in MIGS show task-specific promise but face challenges in clinical translation.
• Current evidence is largely observational, with significant bias and insufficient validation.
• Multicenter prospective validation is essential before routine clinical adoption of AI in MIGS.