Medical Counselling in Imminent Premature Birth: The Chatbot as Expert And Friend? A Pilot Study.

Abstract

Introduction: Medical counselling and shared decision-making in emotionally demanding situations, such as prenatal counselling at the limits of viability, requires clear, empathetic, and context-sensitive communication. Increasingly, patients and families seek medical information from artificial intelligence (AI)-based tools such as ChatGPT. This study evaluated the potential role and limitations of ChatGPT in prenatal counselling for imminent extremely preterm birth.

Methods: ChatGPT-generated responses to prenatal counselling prompts were evaluated by a group of neonatologists. The responses were rated across predefined domains, including medical accuracy and empathy. In addition, qualitative assessments were performed to identify recurring strengths, limitations, and clinically relevant concerns in the generated responses.

Results: ChatGPT provided structured and generally understandable information with high ratings for empathy. However, important limitations were identified regarding nuance, contextual sensitivity, and the explicit communication that treatment decisions should be made jointly by parents and the attending medical team. Responses occasionally lacked sufficient emphasis on individualised counselling and the complexity of ethically challenging clinical situations.

Discussion: AI-based tools may become increasingly relevant as supplementary information sources in medical counselling. At the same time, they introduce new challenges and perspectives for clinicians in maintaining trust, shared decision-making, and therapeutic relationships in the context of diverse and potentially conflicting information sources. Proactive engagement with AI-assisted communication may therefore become an important aspect of future counselling practice.