Assessing Empathetic Engagement in AI-Generated Clinical Dialogues: A Computational-Psychological Hybrid Approach

Abstract

This study evaluates the empathetic quality of AI-generated clinical dialogues using a novel hybrid methodology combining computational text analysis and expert psychological assessment. We analyze verbatim transcripts of interactions between DeepSeek AI and role-playing patients, assessed by licensed psychologists against empathy criteria from A Computational Approach to Understanding Empathy Expressed in Text-Based Mental Health Support. Our results demonstrate significant variance in empathetic expression across dialogue segments, highlighting strengths in active listening but limitations in contextual emotional adaptation. This work bridges gaps between AI performance metrics and human-centered care standards in digital health interventions.

Ähnliche Arbeiten