{"id":"W3096461368","doi":"10.2196/18953","title":"The Impact of Pretrained Language Models on Negation and Speculation Detection in Cross-Lingual Medical Text: Comparative Study","year":2020,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":30,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Ministerio de Economía y Competitividad","keywords":"Computer science; Negation; Natural language processing; Artificial intelligence; Speculation; Conditional random field; Programming language","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004623769,0.0001056815,0.0001686581,0.00002948827,0.00005694563,0.00002355875,0.0001529395,0.0002265243,0.00001446453],"category_scores_gemma":[0.0008222687,0.00006435309,0.00004532827,0.0001307963,0.0002621323,0.000009255203,0.00007913232,0.0002486256,0.000001882536],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001737521,"about_ca_system_score_gemma":0.0001133838,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000306951,"about_ca_topic_score_gemma":0.00007832007,"domain_scores_codex":[0.9987376,0.00007787247,0.0004421869,0.00009900112,0.0005006848,0.0001426725],"domain_scores_gemma":[0.999438,0.0001049551,0.0001284812,0.0001155585,0.00004506877,0.0001679112],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.003620479,0.001470237,0.02704931,0.0003092795,0.0005230106,0.00004232307,0.1929313,0.003394135,0.007853336,0.0003308136,0.001292411,0.7611834],"study_design_scores_gemma":[0.005863104,0.00750588,0.08750965,0.0001351849,0.00001882876,0.00002577372,0.03635747,0.8507042,0.01102103,0.0002221691,0.0002672042,0.0003694896],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9971835,0.00006574125,0.001858507,0.0001493426,0.00003169174,0.000280754,0.000004860087,0.0000137839,0.0004118056],"genre_scores_gemma":[0.9996095,0.00002564083,0.00008759593,0.0001329144,0.0000941901,0.00001668252,0.00002258293,0.000004375289,0.000006538849],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8473101,"threshold_uncertainty_score":0.2624243,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03040752701534484,"score_gpt":0.3791577345735764,"score_spread":0.3487502075582316,"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."}}