{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00130826,0.0001609636,0.0004479733,0.0001788777,0.0001110177,0.000006722804,0.00009345853,0.00004299275,0.00001039645],"category_scores_gemma":[0.00001299071,0.0001122528,0.00002762564,0.0003988922,0.00008196758,0.0001688348,0.00008479371,0.0002352089,2.441108e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001188958,"about_ca_system_score_gemma":0.000057305,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01868916,"about_ca_topic_score_gemma":0.005502428,"domain_scores_codex":[0.997017,0.001342276,0.0008036913,0.0002934762,0.0002662321,0.0002773834],"domain_scores_gemma":[0.9988278,0.0005765784,0.0003094689,0.000210625,0.00001294994,0.00006258709],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.0007174856,0.001104339,0.2015444,0.00426056,0.00004528489,0.00004974568,0.7807098,0.00004848507,0.0005329533,0.0006968705,0.0003406278,0.009949493],"study_design_scores_gemma":[0.004908898,0.0094075,0.2526668,0.002190088,0.0000183849,0.0002362673,0.7255178,0.003533053,0.001313412,0.00003684234,0.000003964509,0.0001669605],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9928929,0.003737801,0.000004167841,0.0004042868,0.0001965602,0.00257684,0.0001366191,0.000009955294,0.00004092469],"genre_scores_gemma":[0.9994739,0.00004227176,0.00002647029,0.0002082736,0.00002185862,0.0001857418,0.00002504875,0.00001222635,0.000004217381],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.05519196,"threshold_uncertainty_score":0.9878455,"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."}}