Interpretable personalized surgical recommendation with joint consideration of multiple decisional dimensions

Area of Science:

• Artificial Intelligence
• Surgical Planning
• Medical Informatics

Background:

• Surgical planning involves complex decisions balancing effectiveness, risk, cost, and patient specifics.
• Personalized surgical recommendations are crucial for optimizing patient outcomes and satisfaction.

Purpose of the Study:

• To introduce Multi-Dimensional Recommendation (MUDI), an interpretable, data-driven AI system for personalized surgical recommendations.
• To evaluate MUDI's performance in the context of Pelvic Organ Prolapse (POP) surgery.

Main Methods:

• Developed MUDI, an intelligent system considering multiple decisional dimensions for joint patient and surgeon recommendations.
• Applied MUDI to Pelvic Organ Prolapse cases, comparing its performance against existing methods and expert urogynecologists.

Main Results:

• MUDI demonstrated excellent performance in POP surgical recommendations, comparable to top urogynecologists.
• The system's transparent process facilitated better communication between surgeons and patients.
• Users accepted MUDI's recommendations and achieved higher surgical accuracy with its aid.

Conclusions:

• MUDI offers a novel approach to personalized surgical planning by integrating multiple decision factors.
• The AI system shows significant potential for improving surgical decision-making and patient care in POP and other complex conditions.