{"id":"W3043746161","doi":"10.48550/arxiv.2007.10185","title":"A Comprehensive Evaluation of Multi-task Learning and Multi-task Pre-training on EHR Time-series Data","year":2020,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Task (project management); Modalities; Scalability; Machine learning; Multi-task learning; Context (archaeology); Transfer of learning; Artificial intelligence; Set (abstract data type); Training set; Health records; Health care; Data science; Database","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01686012,0.003328299,0.002608786,0.001652366,0.001552807,0.001530142,0.002475769,0.003211466,0.003430326],"category_scores_gemma":[0.03877882,0.0006527845,0.001721307,0.001670316,0.001168136,0.003339934,0.002806792,0.004215904,0.002164413],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001595189,"about_ca_system_score_gemma":0.003064337,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01372371,"about_ca_topic_score_gemma":0.01635407,"domain_scores_codex":[0.9922354,0.003471768,0.0005985679,0.001778125,0.001375027,0.0005411237],"domain_scores_gemma":[0.9838937,0.01009788,0.0005475999,0.00206348,0.002367501,0.001029808],"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":[0.005303022,0.004596262,0.01281236,0.002542829,0.001359148,0.0004609308,0.0003470109,0.2209126,0.01263438,0.001629975,0.04854583,0.6888555],"study_design_scores_gemma":[0.0004123783,0.002268414,0.01032992,0.000260564,0.0002079812,0.0004155429,0.0003024632,0.9634814,0.009732045,0.003092768,0.009375913,0.0001206752],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.4785021,0.03778039,0.4198113,0.005545229,0.00487564,0.002957545,0.007603069,0.02678957,0.01613519],"genre_scores_gemma":[0.7175487,0.002966297,0.2454458,0.002131507,0.0006386862,0.001030161,0.02205205,0.0009350305,0.007251878],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.01686012,"threshold_uncertainty_score":0.08916593,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2806089308569709,"score_gpt":0.3034654669005989,"score_spread":0.02285653604362797,"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."}}