OpenAI o1 and DeepSeek-R1, evaluated on a new synthetic multi-turn mental-health dialogue benchmark, scored below 'good' on patient-centric communication and only about 31% on exact diagnosis matching.
Table 25 shows the prompt used to evaluate perceived severity
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Reasoning Is Not All You Need: Examining LLMs for Multi-Turn Mental Health Conversations
OpenAI o1 and DeepSeek-R1, evaluated on a new synthetic multi-turn mental-health dialogue benchmark, scored below 'good' on patient-centric communication and only about 31% on exact diagnosis matching.