{"id":"W4415623941","doi":"10.21203/rs.3.rs-7752202/v1","title":"Towards Multimodal Retrieval-Augmented Generation for Medical Visual Question Answering","year":2025,"lang":"","type":"preprint","venue":"Research Square","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"Carleton University","funders":"","keywords":"Task (project management); Question answering; Trustworthiness; Reliability (semiconductor); Natural language generation; Natural language; Health care; Medical information","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":[0.001657334,0.001390247,0.001393889,0.001531782,0.0005914569,0.00215395,0.002151149,0.002991761,0.02076743],"category_scores_gemma":[0.005930577,0.0006150429,0.001981943,0.0008575656,0.0007911127,0.00209164,0.003415063,0.001794052,0.008127565],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008184087,"about_ca_system_score_gemma":0.0009515147,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004461043,"about_ca_topic_score_gemma":0.004824744,"domain_scores_codex":[0.9985592,0.0006170508,0.00007337537,0.0003704313,0.0002252624,0.0001547888],"domain_scores_gemma":[0.9980887,0.0009177174,0.00006261999,0.0004263576,0.0004173808,0.00008732539],"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":[0.001155405,0.000494817,0.0009803887,0.0007625411,0.0001696847,0.0005553126,0.0006130358,0.03113566,0.07929469,0.01650373,0.05467039,0.8136644],"study_design_scores_gemma":[0.0001491387,0.0002638573,0.0008380295,0.00008987795,0.0001309324,0.0003989309,0.0002540377,0.8955763,0.04023068,0.03875509,0.02324854,0.00006457019],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01827169,0.001279625,0.9511071,0.001169987,0.0004493197,0.0004085255,0.002146707,0.01906922,0.006097662],"genre_scores_gemma":[0.2597684,0.0006863634,0.7141058,0.001059186,0.0004112483,0.0005466689,0.008466264,0.00153726,0.01341878],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.02076743,"threshold_uncertainty_score":0.06947392,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06851537538957315,"score_gpt":0.4775223534019001,"score_spread":0.4090069780123269,"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."}}