To assess the potential of AI-based systems, particularly large language models (LLMs), in improving communication with patients before surgical procedures.
Approach:
Introduction: Discusses the importance of effective communication in reducing patient anxiety and enhancing understanding prior to invasive procedures.
Digital Technologies: Explores how digital technologies can improve informed consent processes without increasing anxiety or reducing satisfaction.
LLMs in Healthcare: Examines the potential of LLMs in patient education and the challenges associated with their deployment in clinical settings.
Reliability Concerns: Addresses the risks of misinformation and biases in LLM outputs, highlighting the concept of 'hallucinations' in AI-generated content.
Retrieval-Augmented Generation (RAG): Introduces RAG as a strategy to enhance the reliability of LLMs by referencing external knowledge bases.
Key Findings:
Effective communication is critical for patient engagement and satisfaction.
Digital technologies can enhance informed consent processes while maintaining patient understanding.
LLMs show promise in improving patient education but pose risks of misinformation.
RAG can improve the reliability of LLM outputs but has its own challenges.
Interpretation:
The deployment of AI-based systems in healthcare communication has potential benefits but also significant challenges that need to be addressed.
Limitations:
LLMs may generate factually incorrect content.
Biases in training data can exacerbate health disparities.
Hallucinations are an inherent limitation of LLM technology.
Conclusion:
AI-based systems, particularly LLMs, have the potential to enhance patient communication but require careful implementation to mitigate risks.
An ancillary analysis found less than 2-percentage-point differences in 10-year recurrence across margin thresholds, with no statistically significant adjusted association.
