{"id":"W4417441746","doi":"10.2196/69143","title":"Application of Machine Learning to Auto-Code Injury Data in the e-CHIRPP System: Development and Evaluation Study","year":2025,"lang":"en","type":"article","venue":"Online Journal of Public Health Informatics","topic":"Data-Driven Disease Surveillance","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Public Health Agency of Canada","funders":"","keywords":"Adaptability; Development (topology); Support vector machine; Training set; Active learning (machine learning)","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01607211,0.00009859756,0.0004014745,0.000492448,0.00007845162,0.00003395384,0.0003894312,0.00003098876,0.000002497908],"category_scores_gemma":[0.001770897,0.00006754049,0.0000169438,0.0006279047,0.00001761076,0.0003295601,0.0001808555,0.0003399709,0.000001766365],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003062436,"about_ca_system_score_gemma":0.002432585,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004220716,"about_ca_topic_score_gemma":0.0001844176,"domain_scores_codex":[0.9963895,0.0004118508,0.002071956,0.00008503939,0.0008498028,0.0001918187],"domain_scores_gemma":[0.9976373,0.0001274215,0.00107032,0.0004930813,0.0005086375,0.0001632415],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0001206159,0.001053962,0.2194007,0.002371945,0.0001328098,0.00000212061,0.01482007,0.0003373654,0.000005277157,0.0001315394,0.001399774,0.7602237],"study_design_scores_gemma":[0.002541305,0.000632041,0.5770957,0.0006683433,0.00007060533,0.00004493023,0.02400077,0.297775,0.000001826043,0.000006690232,0.0970755,0.00008731334],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9604769,0.0004641875,0.02922955,0.00763585,0.0001114787,0.001695087,0.0001866247,0.00001832753,0.0001819771],"genre_scores_gemma":[0.9826578,0.00005641874,0.0156854,0.001047274,0.0000412064,0.00001343517,0.0004831326,0.00000592643,0.000009407477],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7601365,"threshold_uncertainty_score":0.5570303,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08985256827463871,"score_gpt":0.4256305703726556,"score_spread":0.3357780020980168,"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."}}