{"id":"W4200578556","doi":"10.2196/35391","title":"Assessing Generalizability of Deep Learning Models Trained on Standardized and Nonstandardized Images and Their Performance Against Teledermatologists","year":2021,"lang":"en","type":"article","venue":"Iproceedings","topic":"Cutaneous Melanoma Detection and Management","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Convolutional neural network; Generalizability theory; Artificial intelligence; Standardization; Computer science; Deep learning; Pattern recognition (psychology); Receiver operating characteristic; Machine learning; Mathematics; Statistics","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01764333,0.001140801,0.0006440249,0.001286797,0.0004798931,0.001291052,0.001038386,0.00151772,0.001783236],"category_scores_gemma":[0.05502843,0.0004109282,0.001302778,0.0006375882,0.00114205,0.001286904,0.00163362,0.001356881,0.0008763904],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001067615,"about_ca_system_score_gemma":0.0005679276,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005917102,"about_ca_topic_score_gemma":0.004249628,"domain_scores_codex":[0.9921686,0.002537183,0.0007886849,0.003041655,0.001097418,0.0003665813],"domain_scores_gemma":[0.9574766,0.02675671,0.003500951,0.006864761,0.004595571,0.0008053765],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.006079159,0.0008590969,0.7284748,0.0006042306,0.003705347,0.000693901,0.001216969,0.09308687,0.01439546,0.0006151287,0.005089413,0.1451796],"study_design_scores_gemma":[0.0004949469,0.005982964,0.605517,0.0002906681,0.001621767,0.001990607,0.001668492,0.3447514,0.02972398,0.002779806,0.004991348,0.0001868725],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9818264,0.001282621,0.01224004,0.000472745,0.0002331184,0.0001832894,0.001168683,0.0003690396,0.002224006],"genre_scores_gemma":[0.993972,0.0001791536,0.003207534,0.0001916887,0.00004924667,0.00008096376,0.001678069,0.00007028446,0.0005709795],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01764333,"threshold_uncertainty_score":0.09330797,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02582374461614296,"score_gpt":0.2677876922085836,"score_spread":0.2419639475924406,"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."}}