A study of ChatGPT in facilitating Heart Team decisions on severe aortic stenosis

Adil Salihu ¹, David Meier ¹, Nathalie Noirclerc ¹

⁰Department of Cardiology, Lausanne University Hospital, University of Lausanne, Lausanne, Switzerland.

Eurointervention : Journal of Europcr in Collaboration with the Working Group on Interventional Cardiology of the European Society of Cardiology +

|

April 17, 2024

View abstract on PubMed

Summary

Artificial intelligence (AI), specifically large language models (LLMs), shows promise in supporting multidisciplinary Heart Teams (HTs) for valvular heart disease management. ChatGPT achieved 77% agreement with HT decisions, aiding clinical decision-making.

Area Of Science

Cardiology
Artificial Intelligence in Medicine
Medical Informatics

Background

Multidisciplinary Heart Teams (HTs) are crucial for managing valvular heart diseases.
Logistical challenges can impede comprehensive patient data evaluation by HTs, potentially impacting care.
Improving HT efficiency and decision-making is essential for optimal patient outcomes.

Purpose Of The Study

To investigate the potential of artificial intelligence (AI), particularly large language models (LLMs), to enhance HT efficiency.
To assess AI's capability in improving clinical decision-making processes within HTs.
To evaluate the consistency of AI-generated recommendations with expert HT decisions.

Main Methods

Retrospective analysis of patient data from severe aortic stenosis cases presented at HT meetings.
Utilizing OpenAI's GPT-4 to process a standardized questionnaire with 14 key variables.
Comparing AI-generated treatment decisions against those made by the human multidisciplinary Heart Team.

Main Results

The study analyzed data from 150 patients.
ChatGPT demonstrated a 77% agreement rate with the HT's final decisions.
Agreement varied by treatment: 90% for transcatheter valve implantation, 65% for surgical valve replacement, and 65% for medical management.

Conclusions

Large language models (LLMs) present a significant opportunity to refine HT decision-making in valvular heart disease.
AI tools like ChatGPT can serve as a valuable failsafe, flagging discrepancies for further review.
Further research is needed to fully integrate AI into HT workflows for enhanced clinical decision support.