{"id":"W4406061941","doi":"10.1016/j.lungcan.2025.108080","title":"Assessing the feasibility and external validity of natural language processing-extracted data for advanced lung cancer patients","year":2025,"lang":"en","type":"article","venue":"Lung Cancer","topic":"Lung Cancer Research Studies","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Roche (Canada); University Health Network; Systems, Applications & Products in Data Processing (Canada); Princess Margaret Cancer Centre","funders":"Princess Margaret Cancer Foundation; F. Hoffmann-La Roche","keywords":"Medicine; Lung cancer; Intensive care medicine; Natural language processing; Medical physics; Oncology; Computer science","routes":{"ca_aff":true,"ca_fund":true,"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":[0.2018125,0.0008226223,0.0009717872,0.004938737,0.001139526,0.003515671,0.001695333,0.001274363,0.001799289],"category_scores_gemma":[0.4618553,0.0005299869,0.00373673,0.004597426,0.003178272,0.00254979,0.004334071,0.001258418,0.0008510266],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001748981,"about_ca_system_score_gemma":0.003454513,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00165142,"about_ca_topic_score_gemma":0.001422425,"domain_scores_codex":[0.8304647,0.1098808,0.02753407,0.011816,0.01876189,0.001542531],"domain_scores_gemma":[0.3434744,0.5134349,0.05579402,0.03973525,0.04583623,0.001725248],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.002891375,0.0003760869,0.92563,0.001208117,0.001439844,0.0001697065,0.002338466,0.003280846,0.0009010384,0.0009398295,0.002591996,0.0582327],"study_design_scores_gemma":[0.0009469618,0.002783861,0.9004292,0.00161597,0.001644592,0.0009559835,0.003277971,0.05865235,0.007533432,0.006697562,0.01523782,0.0002243825],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9186535,0.002142935,0.05642425,0.002645593,0.0005450443,0.004161082,0.008595969,0.0004493051,0.006382385],"genre_scores_gemma":[0.9656526,0.0003297944,0.02352146,0.0005572633,0.0002276772,0.002563635,0.006772868,0.0001036464,0.0002710544],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2018125,"threshold_uncertainty_score":0.9843073,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06031855068028288,"score_gpt":0.4893047198645877,"score_spread":0.4289861691843049,"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."}}