{"id":"W3036632634","doi":"10.2196/18089","title":"Assessment of the Robustness of Convolutional Neural Networks in Labeling Noise by Using Chest X-Ray Images From Multiple Centers","year":2020,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"COVID-19 diagnosis using AI","field":"Medicine","cited_by":18,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Korea Health Industry Development Institute","keywords":"Convolutional neural network; Receiver operating characteristic; Artificial intelligence; Deep learning; Computer science; Robustness (evolution); Test set; Pattern recognition (psychology); Medicine; Binary classification; Computer-aided diagnosis; Random forest; Noise (video); Medical imaging; Machine learning; Radiology; Support vector machine; 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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002761738,0.0001437419,0.0004192324,0.00004949507,0.00003635212,0.00001090194,0.0002490762,0.000152161,0.00009875366],"category_scores_gemma":[0.0006514177,0.000106292,0.000101139,0.0003338273,0.0002669015,0.0001325515,0.0001963111,0.0005206498,5.03263e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001142814,"about_ca_system_score_gemma":0.0003318999,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001773948,"about_ca_topic_score_gemma":0.000009152189,"domain_scores_codex":[0.9978328,0.00006132005,0.0009105122,0.0001022118,0.0008680187,0.0002251188],"domain_scores_gemma":[0.9985543,0.0005534699,0.0003517884,0.0002127642,0.0001094402,0.000218215],"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":[0.000159301,0.000946232,0.4706934,0.001733745,0.0001648779,0.0000142996,0.004015287,0.4986984,0.00506378,0.00001353894,0.01568989,0.00280727],"study_design_scores_gemma":[0.002214554,0.00004944066,0.04692711,0.0007269201,0.00004730805,0.000002079619,0.000707866,0.9485712,0.0003143587,6.218313e-7,0.0003529645,0.00008556349],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9724965,0.00009254634,0.01795012,0.008641218,0.0002433293,0.0004659121,0.00006845737,0.00002554997,0.00001634473],"genre_scores_gemma":[0.9874863,0.00002419284,0.00492021,0.007352635,0.0001193218,0.00001222326,0.00007054961,0.00001317979,0.000001396046],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4498728,"threshold_uncertainty_score":0.4334462,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03116503130725056,"score_gpt":0.3248385462882414,"score_spread":0.2936735149809909,"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."}}