{"id":"W4400641802","doi":"10.2139/ssrn.4895527","title":"High-Quality, Low-Quantity: A Data-Centric Approach to Deep Learning Performance Optimization in Digital X-Ray Radiography","year":2024,"lang":"en","type":"preprint","venue":"SSRN Electronic Journal","topic":"Medical Imaging Techniques and Applications","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"Université Laval","funders":"","keywords":"Radiography; Computer science; Digital radiography; Artificial intelligence; Deep learning; Quality (philosophy); Medical physics; Computer vision; Radiology; Physics; Medicine","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":[0.003279708,0.001238525,0.00142082,0.0006694327,0.0004151412,0.002074712,0.002876599,0.002420033,0.002425969],"category_scores_gemma":[0.01091908,0.001109305,0.0006501665,0.0008820712,0.001719088,0.003385101,0.004030336,0.002767019,0.0003606081],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001381777,"about_ca_system_score_gemma":0.001582515,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00271332,"about_ca_topic_score_gemma":0.002676461,"domain_scores_codex":[0.9988293,0.0004540505,0.00006523276,0.0001851946,0.000361895,0.0001043564],"domain_scores_gemma":[0.9966988,0.001763929,0.0003649985,0.000426748,0.0005507346,0.000194737],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001800179,0.0001166199,0.0007468932,0.0002099733,0.00005788109,0.00004661303,0.00007432008,0.8878442,0.005132049,0.03597868,0.001851502,0.06776132],"study_design_scores_gemma":[0.000006353835,0.00002320858,0.00007386302,0.000006893229,0.00000505248,0.000007338371,0.000003806977,0.9891359,0.0007413235,0.009787752,0.0002048735,0.000003700798],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0123986,0.000470272,0.9850424,0.0007641917,0.00004443186,0.00003024911,0.00006416886,0.000283537,0.0009022612],"genre_scores_gemma":[0.6449447,0.0008567313,0.345976,0.0005349107,0.0003348161,0.0001661954,0.0003043815,0.0005885864,0.00629362],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003279708,"threshold_uncertainty_score":0.01734495,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02296769679080262,"score_gpt":0.3099165179203398,"score_spread":0.2869488211295371,"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."}}