Patient perspectives on AI-based decision support in surgery.

Area of Science:

• Healthcare technology
• Surgical innovation
• Patient perspectives

Background:

• Predictive machine learning (ML) is rapidly advancing in healthcare, particularly for surgical decisions.
• Limited research exists on patient views regarding ML's use throughout the clinical journey.
• This study explores patient perspectives on intra-operative predictive machine learning (IPML) in colorectal surgery (CRS), specifically for predicting colorectal anastomotic leakage (CAL).

Purpose of the Study:

• To investigate patient views on implementing IPML models in CRS.
• To understand patient perceptions of IPML's role in decision-making and clinical information delivery.

Main Methods:

• Qualitative research design utilizing focus groups and semi-structured interviews.
• Descriptive thematic analysis to identify patient themes and attitudes.
• Sample size determined by saturation analysis (n=19 patients).

Main Results:

• Patients generally perceive IPML models positively in CRS.
• Participants view IPML as an advisory tool, with surgeons retaining primary decision-making authority.
• Emphasis placed on personalized communication, quality of life, and surgeon override capabilities.

Conclusions:

• IPML in CRS is well-received when surgeons maintain override authority and transparent documentation.
• Patient trust in surgeons is crucial for IPML acceptance, necessitating clear communication.
• Tailored communication and addressing quality-of-life impacts (anastomosis vs. stoma) are vital.