{"id":"W4401220095","doi":"10.2196/58202","title":"Revised Tool for the Quality Assessment of Diagnostic Accuracy Studies Using AI (QUADAS-AI): Protocol for a Qualitative Study","year":2024,"lang":"en","type":"article","venue":"JMIR Research Protocols","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":48,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"NIHR Imperial Biomedical Research Centre; National Institute for Health and Care Research","keywords":"Preprint; Protocol (science); Quality assessment; Computer science; Quality (philosophy); Data science; Artificial intelligence; Data mining; Medical physics; Medicine; External quality assessment; World Wide Web; Alternative medicine","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":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.01572616,0.0002025946,0.0005346531,0.0002216357,0.0005468507,0.0001753385,0.0002663934,0.00009566934,0.0000569421],"category_scores_gemma":[0.02614703,0.0001253096,0.0002005906,0.0007146088,0.0003217293,0.000273684,0.0001190358,0.0006408604,0.000006839688],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005396471,"about_ca_system_score_gemma":0.002665926,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002510473,"about_ca_topic_score_gemma":0.00005081327,"domain_scores_codex":[0.9948439,0.001672625,0.001287601,0.0005344936,0.001076793,0.0005846138],"domain_scores_gemma":[0.9542252,0.04143146,0.0002042529,0.0006389636,0.003388188,0.0001118692],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.01310323,0.0152649,0.01862653,0.1585878,0.001816218,0.00002969458,0.3856993,0.00008519464,0.005454949,0.02566683,0.1030019,0.2726635],"study_design_scores_gemma":[0.004136053,0.03851939,0.006184008,0.0462638,0.0001706171,0.0000091575,0.4844694,0.01526202,0.0166923,0.04770787,0.3397269,0.0008584379],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"protocol","genre_gemma":"protocol","genre_scores_codex":[0.003831273,0.00001995866,0.00514837,0.007547824,0.00003273168,0.9833134,0.00003573704,0.00005093112,0.00001979525],"genre_scores_gemma":[0.04581593,0.00000140024,0.001107988,0.0001665383,0.0005154901,0.9521235,0.000006451562,0.00003961223,0.0002230656],"genre_candidate":"protocol","genre_consensus":"protocol","teacher_disagreement_score":0.271805,"threshold_uncertainty_score":0.9820561,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.9118854782094401,"score_gpt":0.8248152070492241,"score_spread":0.08707027116021604,"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."}}