{"id":"W3207181646","doi":"10.2196/33049","title":"Differential Biases and Variabilities of Deep Learning–Based Artificial Intelligence and Human Experts in Clinical Diagnosis: Retrospective Cohort and Survey Study","year":2021,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"COVID-19 diagnosis using AI","field":"Medicine","cited_by":14,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Yonsei University College of Medicine; Ministry of Science and ICT, South Korea; National Research Foundation of Korea; Ministry of Science, ICT and Future Planning; Inje University; Yonsei University; National Research Foundation","keywords":"Artificial intelligence; Medicine; Machine learning; Population; Test (biology); Medical diagnosis; Medical physics; Computer science; Pathology","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":[0.001775318,0.0001886213,0.0008078419,0.0001590962,0.00008308231,0.00005263331,0.00007519424,0.0002247077,0.000212613],"category_scores_gemma":[0.01343224,0.0001633535,0.00004957896,0.0002634533,0.000693016,0.0001072176,0.0002232117,0.0005352511,6.997576e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006331892,"about_ca_system_score_gemma":0.0001963861,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000315554,"about_ca_topic_score_gemma":0.000874145,"domain_scores_codex":[0.9969779,0.00044909,0.001389462,0.0002586528,0.0007011491,0.0002237554],"domain_scores_gemma":[0.9948696,0.004136181,0.0002208232,0.0002528298,0.0002204239,0.0003001535],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00008947401,0.00158917,0.9801113,0.0002851354,0.00006317725,0.00004248608,0.00655726,0.000004831905,0.000003626615,0.0000666635,0.00006267496,0.01112421],"study_design_scores_gemma":[0.001000492,0.001155619,0.9782382,0.0006080355,0.00008123424,0.00001100188,0.007185319,0.01122163,0.000198834,0.000108567,0.00004021346,0.000150806],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9981088,0.0001165729,0.0004078336,0.0005796573,0.00009824533,0.0006329798,0.000008244504,0.00003286205,0.00001478464],"genre_scores_gemma":[0.99844,0.000278545,0.0001633945,0.0008850939,0.00006969089,0.0001148619,0.00003065675,0.0000132577,0.000004502256],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01165692,"threshold_uncertainty_score":0.9948781,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08705776469855132,"score_gpt":0.410996291556137,"score_spread":0.3239385268575857,"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."}}