{"id":"W4406256502","doi":"10.1101/2025.01.07.631773","title":"Biased sampling confounds machine learning prediction of antimicrobial resistance","year":2025,"lang":"en","type":"preprint","venue":"bioRxiv (Cold Spring Harbor Laboratory)","topic":"Bacterial Identification and Susceptibility Testing","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Sampling (signal processing); Artificial intelligence; Machine learning; Resistance (ecology); Antimicrobial; Computer science; Statistics; Mathematics; Biology; Microbiology; Ecology; Computer vision","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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0006487202,0.0003233708,0.0003839224,0.0001519847,0.0001617408,0.0001134826,0.0003705614,0.000512217,0.00002893807],"category_scores_gemma":[0.000890301,0.0003954741,0.0001608016,0.000232714,0.0001135409,0.000007872399,0.0003730876,0.0004455205,0.000003753041],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006759357,"about_ca_system_score_gemma":0.0004744878,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005978184,"about_ca_topic_score_gemma":0.00001818761,"domain_scores_codex":[0.9979084,0.000183523,0.0006447062,0.0008222315,0.0001650587,0.000276082],"domain_scores_gemma":[0.9978467,0.00003605358,0.0005090082,0.0008967342,0.0006203389,0.00009116863],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.00007517588,0.00006949907,0.01226362,0.0005870286,0.0000911453,9.645067e-7,0.000004908083,0.0001841239,0.9864628,0.00006318418,0.0001962275,0.000001337096],"study_design_scores_gemma":[0.000534998,0.00003680501,0.0430197,0.0006062409,0.000092604,8.338291e-9,0.00000213454,0.0003536616,0.9420033,7.679981e-7,0.01296309,0.0003866712],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9919637,0.000515824,0.00473886,0.00006801275,0.001143468,0.0004066587,0.0009949546,0.000134134,0.00003440015],"genre_scores_gemma":[0.9940802,0.0002139507,0.005096917,0.00006986492,0.000335662,0.00003268212,0.00002112326,0.00004336319,0.0001062612],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.04445947,"threshold_uncertainty_score":0.9998497,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02115353240114829,"score_gpt":0.2412234600133923,"score_spread":0.220069927612244,"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."}}