{"id":"W4391479818","doi":"10.2196/52462","title":"Automated Category and Trend Analysis of Scientific Articles on Ophthalmology Using Large Language Models: Development and Usability Study","year":2024,"lang":"en","type":"article","venue":"JMIR Formative Research","topic":"Retinal Imaging and Analysis","field":"Medicine","cited_by":16,"is_retracted":false,"has_abstract":true,"ca_institutions":"Columbia College","funders":"National Eye Institute; National Institutes of Health; Research to Prevent Blindness","keywords":"Categorization; Computer science; Set (abstract data type); Data science; Artificial intelligence; Information retrieval; Natural language processing","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.008184251,0.0009132204,0.0006033294,0.008035479,0.0004166224,0.002066457,0.001064967,0.0007164464,0.001665545],"category_scores_gemma":[0.02945263,0.0003128209,0.001205587,0.003131403,0.0002934232,0.002960949,0.0009642113,0.0008702069,0.001076395],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001066481,"about_ca_system_score_gemma":0.002074823,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004642672,"about_ca_topic_score_gemma":0.006347075,"domain_scores_codex":[0.9953852,0.002056298,0.0005661765,0.0007713023,0.001109773,0.0001111812],"domain_scores_gemma":[0.9663486,0.02522841,0.001654204,0.001988805,0.004327655,0.0004522272],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0008400943,0.001002716,0.05132638,0.001397135,0.0005810196,0.000384247,0.001419366,0.0223276,0.02255664,0.001712644,0.01145679,0.8849955],"study_design_scores_gemma":[0.0001391223,0.0006039849,0.02808638,0.0001734123,0.000269869,0.0005332253,0.00100707,0.9308964,0.02288097,0.004366294,0.01093464,0.0001086444],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3679945,0.001618919,0.5736676,0.001141701,0.0001925064,0.001588702,0.01105408,0.04001015,0.002731843],"genre_scores_gemma":[0.4758447,0.0004098452,0.5120978,0.0001324829,0.0000621927,0.0007421781,0.009320997,0.0003253296,0.001064464],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9919645,"threshold_uncertainty_score":0.04328299,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1227048562417544,"score_gpt":0.4740777505475778,"score_spread":0.3513728943058234,"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."}}