{"id":"W4417151762","doi":"10.64898/2025.11.30.25341311","title":"A Chatbot for the Management of Bipolar Disorder: Using Retrieval-Augmented Generation with an Open-Weight Large Language Model to Answer Clinical Questions Based on the CANMAT and ISBD 2018 Guidelines for Bipolar Disorder","year":2025,"lang":"","type":"article","venue":"medRxiv","topic":"Digital Mental Health Interventions","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto; University of Ottawa; McGill University; University of British Columbia","funders":"","keywords":"Chatbot; Guideline; Consistency (knowledge bases); Logistic regression; Bipolar disorder; Language model; Base (topology)","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007336629,0.001041177,0.0004747427,0.000598832,0.0004918266,0.001139461,0.001326489,0.001350617,0.008774496],"category_scores_gemma":[0.02811807,0.0003774603,0.0005319699,0.0001892857,0.0005205354,0.00132987,0.001734929,0.0009173776,0.002652268],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000715544,"about_ca_system_score_gemma":0.0009071783,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001562823,"about_ca_topic_score_gemma":0.002356088,"domain_scores_codex":[0.9951852,0.003760022,0.0002106674,0.0004370357,0.000276211,0.0001308233],"domain_scores_gemma":[0.9701803,0.02583563,0.0008959809,0.001083231,0.001260832,0.0007440016],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00968066,0.005839617,0.02654299,0.003757738,0.0003448179,0.00380007,0.02854529,0.04213287,0.1003111,0.005555688,0.06405314,0.7094361],"study_design_scores_gemma":[0.002824082,0.01021767,0.01950064,0.001015485,0.0007241027,0.002479802,0.0078438,0.8237781,0.04444509,0.01110861,0.07529178,0.0007709346],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6149077,0.00052641,0.3314136,0.00318397,0.0003859168,0.003812572,0.001486936,0.03334036,0.01094247],"genre_scores_gemma":[0.686883,0.0001351174,0.3029861,0.001242144,0.00008911044,0.001818029,0.001407918,0.0006042526,0.004834402],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.008774496,"threshold_uncertainty_score":0.0388003,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1379372510060407,"score_gpt":0.469319645941257,"score_spread":0.3313823949352163,"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."}}