{"id":"W4404665724","doi":"10.2196/57986","title":"Exploring Biases of Large Language Models in the Field of Mental Health: Comparative Questionnaire Study of the Effect of Gender and Sexual Orientation in Anorexia Nervosa and Bulimia Nervosa Case Vignettes","year":2024,"lang":"en","type":"article","venue":"JMIR Mental Health","topic":"Mental Health via Writing","field":"Psychology","cited_by":18,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Preprint; Upload; 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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.06587515,0.0003756805,0.0006585091,0.002553406,0.001446615,0.00137468,0.001037294,0.00085533,0.002221767],"category_scores_gemma":[0.2321103,0.0004664015,0.001463335,0.001969751,0.002268084,0.00168075,0.002857345,0.0009091106,0.0002411318],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.005064362,"about_ca_system_score_gemma":0.002222671,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002364317,"about_ca_topic_score_gemma":0.003134675,"domain_scores_codex":[0.9115268,0.0759965,0.005458653,0.001649234,0.004359976,0.001008875],"domain_scores_gemma":[0.7708615,0.1876137,0.02133906,0.007189574,0.01123259,0.001763612],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.004358865,0.004962255,0.4754788,0.005479949,0.001007941,0.003647053,0.3636555,0.002387984,0.003432224,0.005116282,0.004559227,0.1259139],"study_design_scores_gemma":[0.001098448,0.008773405,0.7229878,0.003875348,0.000787655,0.004812762,0.2207555,0.006984929,0.003609969,0.003946799,0.02183916,0.0005281935],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9951361,0.0004214249,0.001950824,0.0003667783,0.00002854556,0.001223063,0.0001643238,0.000004989012,0.0007039615],"genre_scores_gemma":[0.9909897,0.0003186956,0.004447826,0.0003187092,0.00002156459,0.003481115,0.0002456039,0.000006389695,0.0001703278],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.06587515,"threshold_uncertainty_score":0.3483853,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1231035052776538,"score_gpt":0.463358199679422,"score_spread":0.3402546944017681,"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."}}