Expert-Level Crisis Detection in Mental Health Conversations

• URL: http://arxiv.org/abs/2606.10380v1
• Date: Tue, 09 Jun 2026 03:42:14 GMT
• Title: Expert-Level Crisis Detection in Mental Health Conversations
• Abstract summary: We introduce CRADLE-Dialogue, a clinician-annotated benchmark for turn-level crisis detection in conversational settings.<n>The dataset features 600 dialogues with multi-label annotations across clinically grounded risks.<n>We propose an Alert-Confirm evaluation protocol that distinguishes early warning signals (Alert) from turns where a specific crisis becomes explicitly identifiable (Confirm)
• Score: 7.830103991641948
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Real-world crisis intervention is inherently conversational, yet existing research largely focuses on static texts.Real-world crisis intervention is inherently conversational, yet existing research largely focuses on static texts. When applied to multi-turn dialogues, current models exhibit significant performance degradation, struggling to track risk signals that emerge as context evolves. To address this gap, we introduce CRADLE-Dialogue, a clinician-annotated benchmark for turn-level crisis detection in conversational settings. The dataset features 600 dialogues with multi-label annotations across clinically grounded risks, including suicide ideation, self-harm, and child abuse, distinguishing past from ongoing risk. We further propose an Alert-Confirm evaluation protocol that distinguishes early warning signals (Alert) from turns where a specific crisis becomes explicitly identifiable (Confirm), reflecting the clinical need to intervene before risk becomes explicit. Experiments show that identifying when risk emerges is much harder than recognizing that it exists: models achieve only mid-40% to high-60% Micro F1. Additionally, we release a synthetic training corpus and a 32B-parameter model that substantially outperforms existing open-source models and achieves competitive or superior results against proprietary models across turn-level, dialogue-level, and confirm-only evaluation settings.

Related papers

• Risk Governance for Generative AI Mental Health Support: A Multi-Turn Safety Architecture [13.319144217202227]
  Large language models (LLMs) are increasingly used for emotional support despite lacking mechanisms to safely govern evolving mental health risk.<n>We developed a model-agnostic safety governance architecture that combines contextual risk detection, reasoning-based verification, and protocol-guided response generation for multi-turn mental health interactions.
  arXiv  Detail & Related papers  (2026-07-17T14:33:11Z)
• Reverse Probing: Supervised Token-level Uncertainty Quantification for Large Language Models in Clinical Text [4.737321357710923]
  We propose Reverse Probing, which estimates token-level uncertainty directly from pre-existing labeled summaries.<n>Reverse Probing treats the text as a probe into the model's internal state, extracting uncertainty signals from four categories of internal activations.<n>We evaluate on two expert-annotated clinical datasets and outperform eight adapted baselines on all metrics.
  arXiv  Detail & Related papers  (2026-05-27T17:01:04Z)
• CRADLE Bench: A Clinician-Annotated Benchmark for Multi-Faceted Mental Health Crisis and Safety Risk Detection [8.296902072126182]
  We introduce CRADLE BENCH, a benchmark for multi-faceted crisis detection.<n>Our benchmark provides 600 clinician-annotated evaluation examples and 420 development examples.<n>We further fine-tune six crisis detection models on subsets defined by consensus and unanimous ensemble agreement.
  arXiv  Detail & Related papers  (2025-10-27T20:32:38Z)
• Shallow Robustness, Deep Vulnerabilities: Multi-Turn Evaluation of Medical LLMs [9.291589998223696]
  We introduce MedQA-Followup, a framework for evaluating multi-turn robustness in medical question answering.<n>Using controlled interventions on the MedQA dataset, we evaluate five state-of-the-art LLMs.<n>We find that while models perform reasonably well under shallow perturbations, they exhibit severe vulnerabilities in multi-turn settings.
  arXiv  Detail & Related papers  (2025-10-14T08:04:18Z)
• Between Help and Harm: An Evaluation of Mental Health Crisis Handling by LLMs [6.0460961868478975]
  We introduce a unified taxonomy of six clinically-informed mental health crisis categories.<n>We benchmark three state-of-the-art LLMs for their ability to classify crisis types and generate safe, appropriate responses.<n>We identify systemic weaknesses in handling indirect or ambiguous risk signals, a reliance on formulaic and inauthentic default replies, and frequent misalignment with user context.
  arXiv  Detail & Related papers  (2025-09-29T14:42:23Z)
• RADAR: A Risk-Aware Dynamic Multi-Agent Framework for LLM Safety Evaluation via Role-Specialized Collaboration [81.38705556267917]
  Existing safety evaluation methods for large language models (LLMs) suffer from inherent limitations.<n>We introduce a theoretical framework that reconstructs the underlying risk concept space.<n>We propose RADAR, a multi-agent collaborative evaluation framework.
  arXiv  Detail & Related papers  (2025-09-28T09:35:32Z)
• A Risk Ontology for Evaluating AI-Powered Psychotherapy Virtual Agents [13.721977133773192]
  Large Language Models (LLMs) and Intelligent Virtual Agents acting as psychotherapists present opportunities for expanding mental healthcare access.<n>Their deployment has also been linked to serious adverse outcomes, including user harm and suicide.<n>We introduce a novel risk ontology specifically designed for the systematic evaluation of conversational AI psychotherapists.
  arXiv  Detail & Related papers  (2025-05-21T05:01:39Z)
• 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)
• Towards Probing Speech-Specific Risks in Large Multimodal Models: A Taxonomy, Benchmark, and Insights [50.89022445197919]
  We propose a speech-specific risk taxonomy, covering 8 risk categories under hostility (malicious sarcasm and threats), malicious imitation (age, gender, ethnicity), and stereotypical biases (age, gender, ethnicity) Based on the taxonomy, we create a small-scale dataset for evaluating current LMMs capability in detecting these categories of risk.
  arXiv  Detail & Related papers  (2024-06-25T10:08:45Z)
• Non-Invasive Suicide Risk Prediction Through Speech Analysis [74.8396086718266]
  We present a non-invasive, speech-based approach for automatic suicide risk assessment. We extract three sets of features, including wav2vec, interpretable speech and acoustic features, and deep learning-based spectral representations. Our most effective speech model achieves a balanced accuracy of $66.2,%$.
  arXiv  Detail & Related papers  (2024-04-18T12:33:57Z)
• Improving Multiple Sclerosis Lesion Segmentation Across Clinical Sites: A Federated Learning Approach with Noise-Resilient Training [75.40980802817349]
  Deep learning models have shown promise for automatically segmenting MS lesions, but the scarcity of accurately annotated data hinders progress in this area. We introduce a Decoupled Hard Label Correction (DHLC) strategy that considers the imbalanced distribution and fuzzy boundaries of MS lesions. We also introduce a Centrally Enhanced Label Correction (CELC) strategy, which leverages the aggregated central model as a correction teacher for all sites.
  arXiv  Detail & Related papers  (2023-08-31T00:36:10Z)
• Epidemic mitigation by statistical inference from contact tracing data [61.04165571425021]
  We develop Bayesian inference methods to estimate the risk that an individual is infected. We propose to use probabilistic risk estimation in order to optimize testing and quarantining strategies for the control of an epidemic. Our approaches translate into fully distributed algorithms that only require communication between individuals who have recently been in contact.
  arXiv  Detail & Related papers  (2020-09-20T12:24:45Z)

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.