{"id":"W4416754409","doi":"10.1101/2025.11.25.25341010","title":"Development and validation of Retrieval Augmented Generation (RAG) and GraphRAG for complex clinical cases","year":2025,"lang":"","type":"preprint","venue":"medRxiv","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Guideline; Nice; Correctness; Baseline (sea); Graph; Scope (computer science); Clinical trial; Pipeline (software); Health informatics","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":[0.01591681,0.001880271,0.0008317113,0.002924677,0.0005349212,0.002629989,0.002931007,0.002645949,0.007000901],"category_scores_gemma":[0.08990055,0.0006415839,0.001982355,0.001166889,0.001183209,0.002563786,0.003162705,0.002288008,0.002545729],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001580687,"about_ca_system_score_gemma":0.003945735,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008033077,"about_ca_topic_score_gemma":0.01002794,"domain_scores_codex":[0.9887489,0.006635791,0.001005145,0.00185824,0.001493214,0.0002585712],"domain_scores_gemma":[0.941731,0.04495707,0.002255329,0.005981982,0.004112966,0.0009616085],"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":[0.00250748,0.00142204,0.02793791,0.005801783,0.0009207281,0.001736641,0.002449867,0.2078047,0.01266395,0.009428561,0.07324161,0.6540847],"study_design_scores_gemma":[0.001227148,0.001431753,0.006248348,0.0007842256,0.000400768,0.0009671693,0.0006017493,0.9199696,0.0181084,0.01473329,0.03532827,0.0001991387],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2299307,0.006568277,0.5595821,0.006244428,0.001129906,0.006806444,0.02172289,0.1563654,0.01164981],"genre_scores_gemma":[0.3834881,0.0007233479,0.588114,0.001436361,0.0001243526,0.001279209,0.02120582,0.001688111,0.001940668],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01591681,"threshold_uncertainty_score":0.08417714,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2040895321409655,"score_gpt":0.4182530610921112,"score_spread":0.2141635289511457,"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."}}