{"id":"W4393529891","doi":"10.1001/jamanetworkopen.2024.4630","title":"Quality of Large Language Model Responses to Radiation Oncology Patient Care Questions","year":2024,"lang":"en","type":"article","venue":"JAMA Network Open","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":105,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"National Cancer Institute; Canadian Institutes of Health Research","keywords":"Readability; Likert scale; Correctness; Radiation oncology; Harm; Medicine; Computer science; Psychology; Medical education; Radiation therapy; Internal medicine; Social psychology","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009123309,0.0000699537,0.00020421,0.00007192343,0.00008361689,0.00004309155,0.0001028754,0.0001148258,0.0001072974],"category_scores_gemma":[0.0005033059,0.00006296134,0.00004641235,0.0003115534,0.00001641982,0.0001204254,0.00008010476,0.0001614604,0.00007457574],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002619824,"about_ca_system_score_gemma":0.0007103798,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002051966,"about_ca_topic_score_gemma":0.0008945421,"domain_scores_codex":[0.9988028,0.0002444609,0.0004191588,0.0001989998,0.0001382815,0.0001963184],"domain_scores_gemma":[0.9990176,0.0003634835,0.00006561416,0.0002468929,0.0001851916,0.0001212133],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.001400525,0.0001876455,0.01509772,0.0002850068,0.00004424504,0.00001778628,0.08613687,0.006671093,0.0006110673,0.01056972,0.04920982,0.8297685],"study_design_scores_gemma":[0.0008043997,0.00807755,0.1334805,0.005097588,0.0003746518,0.00005183134,0.156319,0.1532497,0.01905172,0.01152543,0.5107865,0.001181048],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9772047,0.003117933,0.002453005,0.009741889,0.0006180206,0.0009466005,0.0000586022,0.00004716951,0.005812127],"genre_scores_gemma":[0.9935864,0.0001073284,0.002910813,0.001894882,0.0004998387,0.0001148989,0.0001129889,0.00001307413,0.0007598175],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8285875,"threshold_uncertainty_score":0.3101972,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1616338542101096,"score_gpt":0.5333206311682803,"score_spread":0.3716867769581708,"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."}}