{"id":"W7131129852","doi":"10.1109/iccvw69036.2025.00142","title":"Prompt4Trust: A Reinforcement Learning Prompt Augmentation Framework for Clinically-Aligned Confidence Calibration in Multimodal Large Language Models","year":2025,"lang":"","type":"article","venue":"","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Mila - Quebec Artificial Intelligence Institute","funders":"Canada First Research Excellence Fund; Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Calibration; Task (project management); Reinforcement learning; Generalization; Key (lock); Codebase; Scalability; Sensitivity (control systems)","routes":{"ca_aff":true,"ca_fund":true,"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":[0.003942539,0.001447711,0.0009856495,0.0004621413,0.0003824864,0.001035578,0.002077833,0.001742566,0.005590869],"category_scores_gemma":[0.02250444,0.0005782791,0.0009654146,0.0002571525,0.001054835,0.001984026,0.002918375,0.003678203,0.001677199],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001018366,"about_ca_system_score_gemma":0.001969655,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002558709,"about_ca_topic_score_gemma":0.004044407,"domain_scores_codex":[0.9976385,0.001272892,0.0001182226,0.0005082121,0.0003308237,0.0001314173],"domain_scores_gemma":[0.9933457,0.004908061,0.0003573656,0.0004655244,0.0006720982,0.0002511869],"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":[0.001017203,0.0004499248,0.003096096,0.0006305935,0.0001440925,0.0004321343,0.001249639,0.3749731,0.0248954,0.01218339,0.01985163,0.5610768],"study_design_scores_gemma":[0.00006552166,0.0001766964,0.0002194145,0.00003529462,0.00002181733,0.0000688163,0.00005221202,0.9801013,0.004068563,0.01246495,0.002694204,0.00003115199],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01360079,0.0005242756,0.9736325,0.0006809902,0.0001255836,0.0002263486,0.0002768308,0.009669417,0.001263314],"genre_scores_gemma":[0.5609133,0.0003650421,0.4284826,0.001880993,0.0002021209,0.0008608669,0.001177951,0.0009827696,0.005134467],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005590869,"threshold_uncertainty_score":0.02085036,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02777528130868037,"score_gpt":0.381475785817288,"score_spread":0.3537005045086077,"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."}}