{"id":"W4407677752","doi":"10.2196/70014","title":"Evaluating Generative AI in Mental Health: Systematic Review of Capabilities and Limitations","year":2025,"lang":"en","type":"review","venue":"JMIR Mental Health","topic":"Digital Mental Health Interventions","field":"Psychology","cited_by":41,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Preprint; Mental health; Psychology; Artificial intelligence; Computer science; Psychiatry; 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":[0.002697337,0.00057531,0.004217891,0.0005747199,0.0002638334,0.00003280978,0.0003165881,0.000171152,0.0000892904],"category_scores_gemma":[0.0002366668,0.000517147,0.0005044783,0.0007577809,0.0001658267,0.000165541,0.000188709,0.0005504782,0.00006788509],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002305445,"about_ca_system_score_gemma":0.001546987,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000503757,"about_ca_topic_score_gemma":0.0006341113,"domain_scores_codex":[0.9901144,0.003457436,0.004473907,0.0007646322,0.0004862603,0.0007033447],"domain_scores_gemma":[0.9960205,0.0008720584,0.00209239,0.0006377234,0.00007890505,0.0002984553],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","study_design_scores_codex":[0.000005967219,0.000430092,7.538825e-7,0.7285528,0.00008530152,7.105701e-7,0.002287034,1.212367e-8,7.669948e-9,0.002487584,0.005456135,0.2606936],"study_design_scores_gemma":[0.0004333926,0.001310633,0.000009038511,0.964226,0.0001139417,0.00005933137,0.001735606,0.000003539777,7.557805e-8,0.0001450251,0.03170557,0.000257853],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.000009009975,0.9711186,0.000002906707,0.004897039,0.001194863,0.01948396,0.001729614,0.000043432,0.001520532],"genre_scores_gemma":[0.00003182785,0.9837943,0.0003646299,0.004266982,0.00003392619,0.0067678,0.00130546,0.00005439415,0.003380642],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.2604357,"threshold_uncertainty_score":0.999728,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2092849492106648,"score_gpt":0.5638544977107439,"score_spread":0.3545695485000791,"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."}}