Detecting Stealth Sycophancy in Mental-Health Dialogue with Dynamic Emotional Signature Graphs

• URL: http://arxiv.org/abs/2605.03472v1
• Date: Tue, 05 May 2026 07:56:20 GMT
• Title: Detecting Stealth Sycophancy in Mental-Health Dialogue with Dynamic Emotional Signature Graphs
• Abstract summary: We study multi-domain support-dialogue evaluation without relying on large language models as final judges.<n>We find that direct LLM judges and symmetric text-similarity metrics are poorly aligned with therapeutic quality.<n>To address this issue, we propose Dynamic Emotional Signature Graphs (DESG)<n>DESG represents dialogue windows with decoupled clinical states and scores them using asymmetric clinical geometry.
• Score: 4.002154823241671
• License: http://creativecommons.org/publicdomain/zero/1.0/
• Abstract: As conversational AI therapists are increasingly used in psychological support settings, reliable offline evaluation of therapeutic response quality remains an open problem. This paper studies multi-domain support-dialogue evaluation without relying on large language models as final judges. We use a direct LLM judge as a baseline that reads raw dialogue text and predicts whether the target response is harmful, productive, or neutral. We find that direct LLM judges and symmetric text-similarity metrics are poorly aligned with therapeutic quality because the target label depends on clinical direction: whether the response moves the user state toward regulation or reframing, leaves it broadly unchanged, or reinforces deterioration through higher risk affect or cognitive-distortion mass. To address this issue, we propose Dynamic Emotional Signature Graphs (DESG), a model-agnostic evaluator that represents dialogue windows with decoupled clinical states and scores them using asymmetric clinical geometry. We evaluate DESG on a constructed diagnostic stress-test benchmark of 3{,}000 dialogue windows from EmpatheticDialogues, ESConv, and CRADLE-Dialogue, covering peer support, counseling dialogue, and crisis-oriented interaction. On the 600-window held-out test aggregate, DESG-Ensemble achieves 0.9353 macro-F1, exceeding ConcatANN by 1.51 percentage points, BERTScore by 19.63 points, and TRACT by 33.81 points. Feature ablations, artifact controls, a 100-window blinded adjudicator audit, and qualitative disagreement cases indicate that the clinical state manifold is the main discriminative substrate, while graph-based trajectory components provide asymmetric scoring and interpretable diagnostics rather than serving as the sole source of performance.

Related papers

• Clinician-Level Agreement Without Clinical Caution: LLM Evaluator Limits in Medical AI Benchmarking [27.829988205900055]
  We introduce MedQADE, the first standardised open-response clinical benchmark for German.<n>The top-performing evaluator model, Gemini 3 Flash, reached alignment consistent with the physician ceiling.<n>Despite this statistical alignment, automated evaluators exhibited near-absent clinical metacognition.
  arXiv  Detail & Related papers  (2026-07-01T15:55:31Z)
• Training Therapeutic Judges and Multi-Agent Systems for Human-Aligned Mental Health Support [22.031194303408792]
  We introduce a framework that formulates therapeutic response generation as a decision-refinement problem driven by human-aligned evaluation.<n>In Stage I, we introduce TheraJudge, an open-source therapeutic evaluator trained via preference-based optimization on human-annotated data.<n>In Stage II, we introduce TheraAgent, which operationalizes TheraJudge's evaluations through a coordinated refinement process.
  arXiv  Detail & Related papers  (2026-06-29T20:22:25Z)
• Analyzing and Improving Fine-grained Preference Optimization in Medical LVLMs [53.697403481143404]
  Large Vision-Language Models (LVLMs) have achieved strong performance across medical imaging tasks.<n>But they remain prone to factual inconsistencies, poor visual grounding, and misalignment with clinically meaningful feedback.
  arXiv  Detail & Related papers  (2026-06-10T18:35:36Z)
• A Proactive Multi-Agent Dialogue Framework for Assessing Social Language Disorder Traits in Autism [11.474807247446526]
  We present TPA (Think, Plan, Ask), a proactive multi-agent dialogue framework applied to the language assessment component of the Autism Diagnostic Observation Schedule Module 4 (ADOS-2)<n>A doctor agent explicitly reasons about which traits remain unobserved before selecting a clinically grounded strategy and generating a targeted question.<n>A patient agent grounded in real ADOS-2 clinical data enables reproducible evaluation without real patient participation, validated across three independent experiments confirming adequate fidelity to real patient language.
  arXiv  Detail & Related papers  (2026-05-21T19:45:46Z)
• ADAPTS: Agentic Decomposition for Automated Protocol-agnostic Tracking of Symptoms [36.94429692322632]
  ADAPTS (Agentic Decomposition for Automated Protocol-agnostic Tracking of Symptoms) is a framework for automated rating of depression and anxiety severity.<n>This approach decomposes long-form clinical interviews into symptom-specific reasoning tasks.<n>Ratings approximated expert benchmarks more closely than original human ratings.
  arXiv  Detail & Related papers  (2026-05-04T23:08:42Z)
• Agentic Cognitive Profiling: Realigning Automated Alzheimer's Disease Detection with Clinical Construct Validity [66.94391219005291]
  This paper proposes Agentic Cognitive Profiling (ACP), an agentic framework that realigns automated screening with clinical protocol logic.<n>Central to our design is decoupling semantic understanding from measurement by delegating all quantification to deterministic function calling.<n>ACP achieves 90.5% score match rate in task examination and 85.3% accuracy in AD prediction, surpassing popular baselines.
  arXiv  Detail & Related papers  (2026-03-18T06:15:35Z)
• Stop Listening to Me! How Multi-turn Conversations Can Degrade Diagnostic Reasoning [12.166175637413637]
  We evaluate 17 large language models (LLMs) across three clinical datasets.<n>We develop a "stick-or-switch" evaluation framework to measure model conviction.<n>Our experiments reveal the conversation tax, where multi-turn interactions consistently degrade performance.
  arXiv  Detail & Related papers  (2026-03-12T00:14:35Z)
• AgentsEval: Clinically Faithful Evaluation of Medical Imaging Reports via Multi-Agent Reasoning [73.50200033931148]
  We introduce AgentsEval, a multi-agent stream reasoning framework that emulates the collaborative diagnostic workflow of radiologists.<n>By dividing the evaluation process into interpretable steps including criteria definition, evidence extraction, alignment, and consistency scoring, AgentsEval provides explicit reasoning traces and structured clinical feedback.<n> Experimental results demonstrate that AgentsEval delivers clinically aligned, semantically faithful, and interpretable evaluations that remain robust under paraphrastic, semantic, and stylistic perturbations.
  arXiv  Detail & Related papers  (2026-01-23T11:59:13Z)
• WER is Unaware: Assessing How ASR Errors Distort Clinical Understanding in Patient Facing Dialogue [3.468314243424983]
  Automatic Speech Recognition (ASR) is increasingly deployed in clinical dialogue.<n>Standard evaluations still rely heavily on Error Error Rate (WER)<n>This paper challenges that standard, investigating whether WER or other common metrics correlate with clinical impact of transcription errors.
  arXiv  Detail & Related papers  (2025-11-20T16:59:20Z)
• Retrieval-Augmented Guardrails for AI-Drafted Patient-Portal Messages: Error Taxonomy Construction and Large-Scale Evaluation [5.555479009357263]
  Asynchronous patient-clinician messaging via EHR portals is a growing source of clinician workload.<n>Our contributions are threefold: (1) we introduce a clinically grounded error ontology comprising 5 domains and 59 granular error codes; (2) we develop a retrieval-augmented evaluation pipeline; and (3) we provide a two-stage prompting architecture using DSPy to enable scalable, interpretable, and hierarchical error detection.
  arXiv  Detail & Related papers  (2025-09-26T16:42:43Z)
• MoodAngels: A Retrieval-augmented Multi-agent Framework for Psychiatry Diagnosis [58.67342568632529]
  MoodAngels is the first specialized multi-agent framework for mood disorder diagnosis.<n>MoodSyn is an open-source dataset of 1,173 synthetic psychiatric cases.
  arXiv  Detail & Related papers  (2025-06-04T09:18:25Z)
• Quantifying the Reasoning Abilities of LLMs on Real-world Clinical Cases [48.87360916431396]
  We introduce MedR-Bench, a benchmarking dataset of 1,453 structured patient cases, annotated with reasoning references.<n>We propose a framework encompassing three critical examination recommendation, diagnostic decision-making, and treatment planning, simulating the entire patient care journey.<n>Using this benchmark, we evaluate five state-of-the-art reasoning LLMs, including DeepSeek-R1, OpenAI-o3-mini, and Gemini-2.0-Flash Thinking, etc.
  arXiv  Detail & Related papers  (2025-03-06T18:35:39Z)
• Dialogue is Better Than Monologue: Instructing Medical LLMs via Strategical Conversations [74.83732294523402]
  We introduce a novel benchmark that simulates real-world diagnostic scenarios, integrating noise and difficulty levels aligned with USMLE standards.<n>We also explore dialogue-based fine-tuning, which transforms static datasets into conversational formats to better capture iterative reasoning processes.<n>Experiments show that dialogue-tuned models outperform traditional methods, with improvements of $9.64%$ in multi-round reasoning scenarios and $6.18%$ in accuracy in a noisy environment.
  arXiv  Detail & Related papers  (2025-01-29T18:58:48Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.