{"id":"W2995438613","doi":"10.48550/arxiv.1912.06174","title":"Training without training data: Improving the generalizability of automated medical abbreviation disambiguation","year":2019,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":5,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Generalizability theory; Computer science; Context (archaeology); Artificial intelligence; Training set; Natural language processing; Labeled data; Machine learning; Representation (politics); Training (meteorology); Scarcity; Psychology","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001341427,0.0002299906,0.000355227,0.00005352167,0.00009685021,0.00002365817,0.001153117,0.0006441436,0.00002601645],"category_scores_gemma":[0.001040509,0.0001902158,0.0001301724,0.0001509318,0.0003794811,0.000009404464,0.001270286,0.0003844421,0.000003529308],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000383316,"about_ca_system_score_gemma":0.0006125912,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001861703,"about_ca_topic_score_gemma":0.00008433349,"domain_scores_codex":[0.9980652,0.0003202774,0.0003235146,0.0008605606,0.000171593,0.0002588043],"domain_scores_gemma":[0.9980183,0.00009209944,0.0003910353,0.001292738,0.000110537,0.00009533698],"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":[0.001418777,0.001021591,0.1548208,0.003544956,0.003457532,0.0001295382,0.01064921,0.2220057,0.08086154,0.005648512,0.007217098,0.5092248],"study_design_scores_gemma":[0.001036424,0.0001751849,0.008721788,0.0002031119,0.0002557745,0.00001002344,0.001982353,0.9833518,0.0008875132,0.000614649,0.002259026,0.0005023794],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8952636,0.0002522123,0.1031241,0.0001528968,0.0004633981,0.0002719371,0.00009370939,0.000106499,0.0002715882],"genre_scores_gemma":[0.9977506,0.0001694587,0.0006830445,0.0001018505,0.0001934293,0.000001403774,0.0009207048,0.00001922486,0.0001603035],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.761346,"threshold_uncertainty_score":0.7756773,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1841075146981282,"score_gpt":0.2660201738176082,"score_spread":0.08191265911947995,"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."}}