{"id":"W4417539435","doi":"10.59275/j.melba.2025-6838","title":"Exploring Fairness and Performance Drivers Across State-of-the-Art Pulmonary Nodule Detection Algorithms","year":2025,"lang":"en","type":"article","venue":"The Journal of Machine Learning for Biomedical Imaging","topic":"Lung Cancer Diagnosis and Treatment","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of British Columbia","funders":"Medical Research Council; Stand Up To Cancer; CRUK Lung Cancer Centre of Excellence; UK Regenerative Medicine Platform; University College London; National Institute for Health and Care Research; Cancer Research UK; LUNGevity Foundation; Wellcome Trust; F. Hoffmann-La Roche; Microsoft Research; University College London Hospitals NHS Foundation Trust; Rosetrees Trust; Roy Castle Lung Cancer Foundation; Gilead Sciences; American Association for Cancer Research; AstraZeneca; GlaxoSmithKline","keywords":"Lung cancer screening; Nodule (geology); Discriminative model; National Lung Screening Trial; Lung cancer; Asymptomatic; Computed tomography; Cancer detection","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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":[0.0008386704,0.0001084666,0.0002378984,0.00009387035,0.0002818513,0.0000165134,0.0001185877,0.00001849225,0.000004045616],"category_scores_gemma":[0.0001408517,0.00005855355,0.0001105396,0.0002338896,0.0002187505,0.000155646,0.00008860783,0.0003961645,6.132209e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000124453,"about_ca_system_score_gemma":0.00005805708,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004518364,"about_ca_topic_score_gemma":0.000003308124,"domain_scores_codex":[0.9990413,0.00006222598,0.0003238086,0.00009859621,0.0002783438,0.0001956864],"domain_scores_gemma":[0.9992511,0.0002079361,0.0002387405,0.0001115745,0.0001173015,0.00007337389],"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":[0.0004819747,0.0001286026,0.1540679,0.0003315262,0.0001973021,0.00001306104,0.0008890523,0.0002675211,0.00342138,0.000002589573,0.00008185546,0.8401173],"study_design_scores_gemma":[0.00825535,0.001555207,0.684148,0.003914193,0.001172821,0.001247269,0.001498606,0.2046442,0.05983278,0.0002087148,0.03324041,0.0002824031],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9812763,0.001154723,0.01145262,0.005295021,0.0005616099,0.0001922318,0.000003609623,0.0000129293,0.00005093382],"genre_scores_gemma":[0.998145,0.00103072,0.0003963503,0.000114015,0.00009810831,0.000009522493,0.000002397719,0.00001237765,0.0001914843],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8398349,"threshold_uncertainty_score":0.2387744,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01740001303978086,"score_gpt":0.2909937764071602,"score_spread":0.2735937633673793,"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."}}