{"paper":{"title":"Expert-Level Crisis Detection in Mental Health Conversations","arxiv_id":"2606.10380","primary_cat":"cs.CL","submitted_at":"2026-06-09T03:42:14+00:00","version_compiled":"v3"},"notice":"Every quoted sentence is the authors' text from the arXiv listing abstract. Labels and probabilities are the outputs of classifier-1.13.0 on the question bank abstract-p1-20260923, exactly as returned. No sentence in this record was written by a language model.","schema":"pith.compiled-record/v0","status":"compiled","abstract":{"source":"arXiv listing abstract v3","sentences":[{"i":0,"p":0.9,"id":"s0","own":false,"role":"background","text":"Real-world crisis intervention is inherently conversational, yet existing research largely focuses on static texts.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.05,"main_claim_p":0.02,"quantitative":false,"quantitative_p":0.14,"role_probabilities":{"gap":0.1,"method":0.0,"result":0.0,"background":0.9,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":1,"p":0.74,"id":"s1","own":false,"role":"gap","text":"When applied to multi-turn dialogues, current models exhibit significant performance degradation, struggling to track risk signals that emerge as context evolves.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.07,"main_claim_p":0.05,"quantitative":false,"quantitative_p":0.12,"role_probabilities":{"gap":0.74,"method":0.0,"result":0.0,"background":0.26,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":2,"p":0.96,"id":"s2","own":true,"role":"contribution","text":"To address this gap, we introduce CRADLE-Dialogue, a clinician-annotated benchmark for turn-level crisis detection in conversational settings.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.92,"main_claim_p":0.06,"quantitative":false,"quantitative_p":0.05,"role_probabilities":{"gap":0.0,"method":0.04,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.96,"interpretation":0.0}},{"i":3,"p":0.54,"id":"s3","own":true,"role":"method","text":"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.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.91,"main_claim_p":0.04,"quantitative":true,"quantitative_p":0.97,"role_probabilities":{"gap":0.0,"method":0.54,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.46,"interpretation":0.0}},{"i":4,"p":0.54,"id":"s4","own":true,"role":"method","text":"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.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.88,"main_claim_p":0.08,"quantitative":false,"quantitative_p":0.06,"role_probabilities":{"gap":0.0,"method":0.54,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.46,"interpretation":0.0}},{"i":5,"p":0.99,"id":"s5","own":true,"role":"result","text":"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.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.93,"main_claim_p":0.32,"quantitative":true,"quantitative_p":0.99,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.99,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.01}},{"i":6,"p":0.66,"id":"s6","own":true,"role":"result","text":"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.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.95,"main_claim_p":0.4,"quantitative":true,"quantitative_p":0.98,"role_probabilities":{"gap":0.0,"method":0.01,"result":0.66,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.33,"interpretation":0.0}}]},"compiled":{"main_claim":{"p":0.61,"text":"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.","location":"abstract s5","hedge_strength":{"p":1.0,"label":"asserted"}},"provenance":{"route":{"via":"TypeSafe API","model":"classifier-1.13.0","endpoint":"https://api.typesafe.ai/v1/systemone","provider":"TypeSafe"},"packing":"one request per paper","requests":1,"input_tokens":5683,"cost_usd_reported":0.0},"compiled_at":"2026-09-23","bank_version":"abstract-p1-20260923","model_version":"classifier-1.13.0","abstract_states":{"claims_first":0.16,"states_limitation":0.64,"human_participants":0.64,"controlled_comparison":0.44,"randomized_assignment":0.01,"states_quantitative_result":0.98,"compares_with_prior_results":0.94,"mentions_code_or_data_availability":0.91},"evidential_basis":{"p":1.0,"value":"observation or experiment","probabilities":{"argument":0.0,"literature":0.0,"proof_or_derivation":0.0,"analysis_of_existing_data":0.0,"observation_or_experiment":1.0,"simulation_or_computation":0.0}},"contribution_type":{"p":1.0,"value":"dataset or benchmark","probabilities":{"review":0.0,"method_or_tool":0.0,"empirical_study":0.0,"simulation_study":0.0,"theoretical_result":0.0,"dataset_or_benchmark":1.0,"position_or_commentary":0.0}},"contribution_statement":{"p":0.95,"text":"To address this gap, we introduce CRADLE-Dialogue, a clinician-annotated benchmark for turn-level crisis detection in conversational settings.","location":"abstract s2"}},"coverage":{"tier":"abstract","levels_run":["P abstract"],"not_examined":["full text","figures and tables","references and citing sentences","relations to other papers"],"sources_read":["arXiv listing abstract v3"],"levels_not_run":["S sections","T body sentences","C citations","R cross-paper relations"],"sentences_examined":7,"sentences_in_abstract":7},"measured_accuracy":{"url":"https://pith.science/accuracy","measured_on":"2026-10-02","method":"two independent model families labelled a fresh random sample blind; the figure is how often the published label matches their consensus","sentence_role_agreement":0.92,"main_claim_agreement":0.908,"typed_citation_agreement":0.933,"typed_citation_agreement_95_interval":[0.917,0.945],"review_comments_accurate":0.758,"review_comments_accurate_95_interval":[0.684,0.82],"review_caught_admitted_error_found_or_related":0.448}}