{"id":"W4405068114","doi":"10.2196/59045","title":"Intersection of Performance, Interpretability, and Fairness in Neural Prototype Tree for Chest X-Ray Pathology Detection: Algorithm Development and Validation Study","year":2024,"lang":"en","type":"article","venue":"JMIR Formative Research","topic":"COVID-19 diagnosis using AI","field":"Medicine","cited_by":5,"is_retracted":false,"has_abstract":true,"ca_institutions":"Health Sciences Centre; Sunnybrook Health Science Centre; St. Michael's Hospital; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; University of Toronto","keywords":"Interpretability; Artificial intelligence; Classifier (UML); Receiver operating characteristic; Machine learning; Decision tree; Computer science; Gradient boosting; Deep learning; Artificial neural network; Pattern recognition (psychology); Random forest","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001809502,0.0001061102,0.0002215725,0.0006728146,0.00009011088,0.00004575316,0.00005056203,0.00006865586,0.000007234652],"category_scores_gemma":[0.0001382668,0.00008761226,0.00002337075,0.0005051821,0.0001566528,0.0003278561,0.0001299494,0.0003441046,0.000002184242],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002795367,"about_ca_system_score_gemma":0.0001384547,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002589997,"about_ca_topic_score_gemma":0.00009558622,"domain_scores_codex":[0.9987103,0.0001866295,0.0003362402,0.0002781178,0.0002772307,0.0002114307],"domain_scores_gemma":[0.9991546,0.0003892242,0.0000355609,0.0001213572,0.0002472362,0.00005200732],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0007026223,0.0005433443,0.02500005,0.002783599,0.00004437845,0.00001046906,0.05440445,0.000005412334,0.001159519,0.00001036151,0.00002537428,0.9153104],"study_design_scores_gemma":[0.00280096,0.01030735,0.7901761,0.0009644771,0.00002756905,0.00006754038,0.01129009,0.150019,0.03202353,0.00009591859,0.002038389,0.0001890694],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.992559,0.00006677557,0.003238236,0.0004697593,0.0001186034,0.003475954,0.000003360917,0.00003220745,0.00003611679],"genre_scores_gemma":[0.9973846,0.00001046038,0.0003605937,0.0000181233,0.00003112564,0.002155044,0.000008131954,0.00001311246,0.00001877144],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9151213,"threshold_uncertainty_score":0.3572724,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06756871581205094,"score_gpt":0.4246695908152568,"score_spread":0.3571008750032059,"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."}}