{"id":"W2802776178","doi":"10.17269/s41997-018-0059-0","title":"Comparing Twitter data to routine data sources in public health surveillance for the 2015 Pan/Parapan American Games: an ecological study","year":2018,"lang":"en","type":"article","venue":"Canadian Journal of Public Health","topic":"Data-Driven Disease Surveillance","field":"Medicine","cited_by":11,"is_retracted":false,"has_abstract":false,"ca_institutions":"McGill University; Toronto Public Health; Kingston Health Sciences Centre; Public Health Ontario; University Health Network; University of Toronto; Queen's University","funders":"","keywords":"Helpline; Social media; Public health surveillance; Medicine; Public health; Big data; Categorization; Computer science; Data mining; World Wide Web; Artificial intelligence","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":true,"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":[0.02056925,0.0002904491,0.001247531,0.0007066197,0.000445379,0.0003528714,0.003299398,0.00006266897,0.00009851595],"category_scores_gemma":[0.004677622,0.0002095927,0.00005775088,0.001238454,0.0004726572,0.0007072418,0.000380331,0.0007229656,0.00001877394],"about_ca_system_candidate":true,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001071352,"about_ca_system_score_gemma":0.01847701,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.06035086,"about_ca_topic_score_gemma":0.7865762,"domain_scores_codex":[0.9938288,0.001567696,0.001421163,0.0007989154,0.0005769526,0.001806479],"domain_scores_gemma":[0.9890034,0.000484378,0.0009569016,0.003127421,0.0006186874,0.005809194],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00005609238,0.0004430871,0.8684957,0.00003520822,0.0001147684,0.00003481761,0.002726104,0.00000419035,3.72135e-7,0.00001823096,0.09602343,0.03204798],"study_design_scores_gemma":[0.001279682,0.002269612,0.6911033,0.00003548637,0.000007643164,0.00007082472,0.007281379,0.001030009,2.786035e-8,0.000003068418,0.296784,0.0001349908],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8251307,0.000791037,0.001571008,0.1690237,0.0004447093,0.001536514,0.001432363,0.00002935765,0.00004062759],"genre_scores_gemma":[0.9752402,0.00006085104,0.001128311,0.02131541,0.001120205,0.00001788582,0.001041847,0.00004469155,0.00003057263],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7262253,"threshold_uncertainty_score":0.9870873,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.3756291943115768,"score_gpt":0.4365406736477606,"score_spread":0.06091147933618379,"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."}}