{"id":"W3116946603","doi":"10.2196/22795","title":"Adapting Bidirectional Encoder Representations from Transformers (BERT) to Assess Clinical Semantic Textual Similarity: Algorithm Development and Validation Study","year":2020,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"Topic Modeling","field":"Computer Science","cited_by":24,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Deutschen Konsortium für Translationale Krebsforschung; Deutsches Krebsforschungszentrum","keywords":"Computer science; Encoder; Pearson product-moment correlation coefficient; Leverage (statistics); Artificial intelligence; Semantic similarity; Natural language processing; Relationship extraction; Sentence; Test set; Ground truth; Machine learning; Information extraction; Statistics","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":[0.0008447204,0.0001446214,0.0002546093,0.00007533843,0.0001765032,0.0001766054,0.000421887,0.0001198044,0.000040531],"category_scores_gemma":[0.0003561679,0.0001319475,0.00004313269,0.0003243605,0.00003868368,0.0006519451,0.0002895344,0.000380381,0.00003683302],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003595816,"about_ca_system_score_gemma":0.0003298765,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003482035,"about_ca_topic_score_gemma":0.00002440433,"domain_scores_codex":[0.9972001,0.0001054582,0.001092892,0.0002680035,0.001105602,0.0002279419],"domain_scores_gemma":[0.9986408,0.0004180803,0.0001254037,0.0001956665,0.00009261835,0.0005273925],"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":[0.000008845202,0.0002460749,0.009660479,0.00003808654,0.0001094146,0.0000180421,0.06629571,0.0003656746,0.000004765808,0.0002888521,0.0006948162,0.9222692],"study_design_scores_gemma":[0.0007704811,0.0001165229,0.007627623,0.0000425567,0.00001566035,0.000006943055,0.007525653,0.9809087,0.0001049606,0.00007202045,0.002602972,0.0002059017],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.270766,0.000003962512,0.7263898,0.001905555,0.0001610577,0.0003785102,0.000003120622,0.0001105388,0.0002813934],"genre_scores_gemma":[0.5044168,0.000006329745,0.4921813,0.003066374,0.000215409,0.00006557904,0.00002409662,0.000008202113,0.00001593474],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.980543,"threshold_uncertainty_score":0.5380661,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1256573497326064,"score_gpt":0.3820119822168946,"score_spread":0.2563546324842882,"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."}}