{"id":"W4416635488","doi":"10.1183/13993003.congress-2025.pa2023","title":"Language Models Lag Behind: Inconsistent Alignment with Recent Removal of Race from Clinical Algorithms","year":2025,"lang":"","type":"article","venue":"","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of British Columbia","funders":"","keywords":"Race (biology); Construct (python library); Function (biology); Renal function; Language model; Lag","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.02996532,0.001983041,0.001084882,0.001426546,0.0008806546,0.003858328,0.002653721,0.001526586,0.00366773],"category_scores_gemma":[0.1402945,0.0008152984,0.0026321,0.001151498,0.0009166884,0.004113204,0.00260671,0.003848488,0.002195113],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002317205,"about_ca_system_score_gemma":0.005668508,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01551117,"about_ca_topic_score_gemma":0.01808077,"domain_scores_codex":[0.9832563,0.009600817,0.001897443,0.003030455,0.00169687,0.0005180153],"domain_scores_gemma":[0.9167698,0.06334269,0.003487116,0.007230826,0.008034584,0.001134877],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.00368169,0.0006647665,0.2046431,0.002727248,0.001755715,0.001395182,0.003528796,0.2851015,0.007825214,0.01056478,0.07844812,0.3996638],"study_design_scores_gemma":[0.0003350977,0.0005293929,0.01748352,0.0007506236,0.0006170117,0.0004038324,0.000786129,0.9176033,0.01083811,0.02171352,0.02875699,0.0001824764],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5406303,0.00561274,0.3608555,0.0209746,0.002645684,0.0005868894,0.01517996,0.04042376,0.01309053],"genre_scores_gemma":[0.8427705,0.000676818,0.1298564,0.00442309,0.0002872983,0.0003133786,0.01717061,0.002399411,0.002102446],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9700347,"threshold_uncertainty_score":0.1584737,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04179178006067401,"score_gpt":0.3593313345883819,"score_spread":0.3175395545277079,"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."}}