Can generative artificial intelligence enhance evidence-based and personalized medicine?

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Clinical Decision Support

Background:

• Generative artificial intelligence (GAI) is increasingly explored for clinical applications.
• Determining the optimal approach for GAI in healthcare—rigid adherence to guidelines versus personalized recommendations—remains a key question.
• Patient-specific factors are crucial in effective medical treatment.

Purpose of the Study:

• To compare clinical treatment recommendations made by generative artificial intelligence (GAI) against those made by physicians.
• To evaluate the impact of context- and patient-specific circumstances on GAI and physician decision-making.
• To inform the design of future GAI clinical applications.

Main Methods:

• A comparative study analyzing treatment recommendations from GAI and human physicians.
• Evaluation of recommendations based on adherence to established treatment guidelines.
• Assessment of the consideration of patient-specific variables in the recommendation process.

Main Results:

• GAI and physician recommendations showed variability in approach.
• The study underscored the significance of tailoring recommendations to individual patient contexts.
• Rigid adherence to guidelines by GAI may not always align with optimal patient care.

Conclusions:

• Clinical applications of GAI should incorporate flexibility, balancing guideline adherence with patient individuality.
• Future GAI tools need to effectively integrate patient-specific data for nuanced decision-making.
• Physician oversight remains vital in interpreting and applying GAI recommendations in complex clinical scenarios.