{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.09163973,0.003010106,0.00279378,0.001368915,0.003293333,0.005540598,0.004835451,0.003481229,0.0142188],"category_scores_gemma":[0.2188256,0.002272414,0.009834791,0.001327857,0.001959838,0.006303668,0.004363281,0.007259374,0.003493657],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003339407,"about_ca_system_score_gemma":0.003413199,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02204967,"about_ca_topic_score_gemma":0.01896587,"domain_scores_codex":[0.9365893,0.05679839,0.001886405,0.003205607,0.001011319,0.0005089057],"domain_scores_gemma":[0.3679126,0.6147233,0.004507897,0.007793598,0.00361659,0.001446099],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"qualitative","study_design_scores_codex":[0.1476826,0.06125169,0.2712499,0.009311074,0.0552469,0.001082435,0.0507111,0.09693378,0.006445588,0.01317131,0.02523097,0.2616827],"study_design_scores_gemma":[0.03452757,0.03734599,0.1911251,0.002118282,0.04926354,0.0008125956,0.01581219,0.6126148,0.006433211,0.03619463,0.01139245,0.002359679],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9529096,0.001171472,0.0371312,0.0009599015,0.0002750878,0.002690977,0.002886737,0.0005890032,0.001385964],"genre_scores_gemma":[0.9317466,0.000377708,0.04525626,0.0005863657,0.000205022,0.0116753,0.006490127,0.0004533745,0.003209264],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.09163973,"threshold_uncertainty_score":0.484643,"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."}}