{"id":"W2961541685","doi":"10.2196/15371","title":"Transfer Learning for Risk Classification of Social Media Posts: Model Evaluation Study","year":2020,"lang":"en","type":"preprint","venue":"Journal of Medical Internet Research","topic":"Mental Health via Writing","field":"Psychology","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Princess Margaret Cancer Centre; University Health Network; University of Toronto; Centre for Addiction and Mental Health","funders":"Centre for Addiction and Mental Health Foundation; Centre for Addiction and Mental Health; Nvidia","keywords":"Computer science; Machine learning; Social media; Artificial intelligence; Triage; Benchmark (surveying); F1 score; Task (project management); Transfer of learning; Natural language processing; World Wide Web; Psychology","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":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.0431808,0.0002001934,0.0007597143,0.0005551907,0.000111409,0.00005250053,0.001283951,0.0007822872,0.00220568],"category_scores_gemma":[0.01076269,0.0001764315,0.0003103576,0.0002230538,0.0002367368,0.00007989012,0.0003736766,0.007610823,0.00002501244],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006037619,"about_ca_system_score_gemma":0.001520456,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002160654,"about_ca_topic_score_gemma":0.00011683,"domain_scores_codex":[0.9846493,0.005426533,0.002089331,0.0004839824,0.006851475,0.0004993843],"domain_scores_gemma":[0.9932202,0.002980401,0.0008013301,0.0002279377,0.002304409,0.0004656878],"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":[0.01255682,0.00666364,0.009758935,0.003063889,0.002635887,0.000184022,0.3428929,0.001225167,0.001256246,0.01121976,0.05500824,0.5535345],"study_design_scores_gemma":[0.01085851,0.004842028,0.03756917,0.001791239,0.0005998287,0.00004899179,0.09668638,0.8348815,0.0002133801,0.01116401,0.0009603513,0.0003845577],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9718741,0.0004963138,0.01632897,0.007257183,0.001244453,0.001671029,0.00003346092,0.00001298484,0.001081538],"genre_scores_gemma":[0.9974614,0.00009489215,0.0002427721,0.00005687476,0.001726925,0.0001977648,0.00004845928,0.00004913765,0.0001217994],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8336564,"threshold_uncertainty_score":0.9987065,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.478219008210874,"score_gpt":0.5996678574571407,"score_spread":0.1214488492462666,"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."}}