{"id":"W4391654261","doi":"10.2196/32690","title":"Vision-Language Model for Generating Textual Descriptions From Clinical Images: Model Development and Validation Study","year":2024,"lang":"en","type":"article","venue":"JMIR Formative Research","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":37,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"China Postdoctoral Science Foundation; National Natural Science Foundation of China","keywords":"Computer science; Natural language processing; Artificial intelligence; Standardization; Deep learning; Metric (unit); Closed captioning; Quality (philosophy); Information retrieval; Machine learning; Image (mathematics)","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003706851,0.001276871,0.0007649247,0.001084704,0.000356091,0.001070211,0.002122273,0.001309059,0.003010946],"category_scores_gemma":[0.009466548,0.000486366,0.001253253,0.0006616924,0.0004346119,0.001170155,0.0009316378,0.002397399,0.001039589],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002035009,"about_ca_system_score_gemma":0.002125803,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03818992,"about_ca_topic_score_gemma":0.02521247,"domain_scores_codex":[0.9989319,0.0003864199,0.00007134265,0.000304092,0.0002000164,0.0001061701],"domain_scores_gemma":[0.9946102,0.003707412,0.0002114609,0.0002951838,0.001077458,0.00009835645],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0006946594,0.0007312707,0.005727993,0.000451925,0.0002228039,0.0002044665,0.000178603,0.7004835,0.004110761,0.001883516,0.005030508,0.28028],"study_design_scores_gemma":[0.00001692871,0.00007261839,0.0003270744,0.000009477271,0.00001695419,0.00001930127,0.00001309036,0.9980299,0.001002331,0.0002561262,0.0002296352,0.000006634823],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.4116251,0.003318192,0.5627276,0.001456818,0.0003781934,0.001317335,0.003026528,0.009001731,0.007148471],"genre_scores_gemma":[0.8157948,0.0006735513,0.1744763,0.0003804725,0.00006146423,0.000818776,0.004240381,0.0001451789,0.003409138],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.03818992,"threshold_uncertainty_score":0.07593524,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1517744827311691,"score_gpt":0.505880174163475,"score_spread":0.3541056914323059,"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."}}