{"id":"W7116741269","doi":"10.2196/82971","title":"Large Language Models for Breast and Cervical Cancer Communication: Mixed-methods Evaluation of Linguistic Quality, Safety, and Accessibility (Preprint)","year":2025,"lang":"en","type":"article","venue":"JMIR Cancer","topic":"Health Literacy and Information Accessibility","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"","funders":"","keywords":"Cervical cancer; Breast cancer; MEDLINE; Pragmatics","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low","status":"direct model label, unvalidated"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high","status":"direct model label, unvalidated"}],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01191573,0.000139498,0.0004009773,0.00008782215,0.0005745924,0.00002264157,0.0002235459,0.0002106193,0.0007454587],"category_scores_gemma":[0.0009862401,0.0001207991,0.00004760235,0.0002396419,0.000103508,0.000447146,0.000390357,0.0003807544,0.000001205819],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003905084,"about_ca_system_score_gemma":0.001106304,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001920957,"about_ca_topic_score_gemma":0.002564447,"domain_scores_codex":[0.9954765,0.002079942,0.00155852,0.0003275317,0.0002587363,0.0002987855],"domain_scores_gemma":[0.9957657,0.001477081,0.0006019963,0.0006549529,0.001387722,0.0001124959],"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":[0.000965017,0.0001914244,0.2436954,0.0102711,0.00007055477,3.073104e-8,0.04999267,0.0001516696,0.0000588169,0.03149099,0.0009242913,0.662188],"study_design_scores_gemma":[0.002744127,0.00001265031,0.8123617,0.0008990543,0.0000667264,2.060894e-7,0.004173907,0.1465236,0.00004693529,0.02634627,0.006668755,0.0001561477],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9058272,0.007520474,0.06472148,0.003831584,0.0008248123,0.005682897,0.001842431,0.00009560528,0.009653507],"genre_scores_gemma":[0.9908513,0.0004721251,0.004290175,0.001797716,0.00009068585,0.002172071,0.00006045552,0.000009627378,0.0002558684],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6620319,"threshold_uncertainty_score":0.816225,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1006760896343456,"score_gpt":0.5973600700214101,"score_spread":0.4966839803870645,"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."}}