{"id":"W4399673257","doi":"10.2196/58129","title":"Large Language Models Versus Expert Clinicians in Crisis Prediction Among Telemental Health Patients: Comparative Study","year":2024,"lang":"en","type":"article","venue":"JMIR Mental Health","topic":"Mental Health via Writing","field":"Psychology","cited_by":32,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Preprint; Mental health; Psychology; Medicine; Psychiatry; Computer science; World Wide Web","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.001482991,0.0003645406,0.0006322621,0.0003794013,0.0004290418,0.00005565445,0.0002200663,0.0001150346,0.0005214947],"category_scores_gemma":[0.000003535243,0.0003765753,0.0001038776,0.0005695628,0.00005083313,0.0005405053,0.000135418,0.0006745234,0.0002290617],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002746462,"about_ca_system_score_gemma":0.0002100118,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01024929,"about_ca_topic_score_gemma":0.007617859,"domain_scores_codex":[0.9946311,0.001070386,0.001495076,0.0009619306,0.0006086786,0.001232885],"domain_scores_gemma":[0.9987085,0.0001337705,0.0002859941,0.0004052368,0.00002447807,0.0004419896],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.001410942,0.01005295,0.09178475,0.0004504539,0.0001422972,0.00005237188,0.7902415,0.000007015242,0.000001771541,0.001652527,0.08392644,0.02027703],"study_design_scores_gemma":[0.01072979,0.007706852,0.1312708,0.0005548227,0.000006747814,0.000003757024,0.8442683,0.002602526,0.00000375766,0.00003985605,0.002463133,0.0003497156],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9815991,0.00314728,0.0000467336,0.001354263,0.004014804,0.005943425,0.001117073,0.0003305278,0.002446815],"genre_scores_gemma":[0.9952896,0.00009260826,0.00009753572,0.002071121,0.0002862422,0.001095842,0.0007722543,0.00006862504,0.0002261707],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.08146331,"threshold_uncertainty_score":0.9998686,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08244833402517719,"score_gpt":0.4919045461106609,"score_spread":0.4094562120854837,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}