Chatbot assistance in precision oncology treatment decision-making

Area of Science:

• Oncology
• Bioinformatics
• Medical Artificial Intelligence

Background:

• Artificial intelligence (AI) chatbots show potential in healthcare applications.
• The efficacy of AI chatbots in interpreting complex molecular data for patient care remains largely unevaluated.

Purpose of the Study:

• To assess the accuracy and comprehensiveness of AI chatbot recommendations using combined clinical and molecular data.
• To identify limitations in AI chatbot interpretation of complex genomic profiles in patient scenarios.

Main Methods:

• Evaluation of 50 distinct patient cases integrating clinical and molecular data.
• Analysis of AI chatbot-generated treatment recommendations for accuracy and completeness.

Main Results:

• AI chatbots provided generally accurate and comprehensive recommendations across evaluated patient scenarios.
• Occasional omissions of critical treatment options were noted.
• In cases with multiple mutations, chatbots sometimes suggested non-data-driven treatment options.

Conclusions:

• AI chatbots demonstrate considerable capability in processing molecular data for medical recommendations.
• Further refinement is needed to ensure complete inclusion of all relevant treatment options and avoid non-data-driven suggestions, especially in complex genomic cases.