{"id":"W2965865809","doi":"10.1093/jamia/ocz114","title":"The machine giveth and the machine taketh away: a parrot attack on clinical text deidentified with hiding in plain sight","year":2019,"lang":"en","type":"article","venue":"Journal of the American Medical Informatics Association","topic":"Electronic Health Records Systems","field":"Health Professions","cited_by":12,"is_retracted":false,"has_abstract":true,"ca_institutions":"Privacy Analytics (Canada)","funders":"U.S. National Library of Medicine; National Human Genome Research Institute","keywords":"Computer science; Identifier; Sight; Artificial intelligence; Masking (illustration); Natural language processing; Random forest; Computer security; Information retrieval","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":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.008773603,0.001156204,0.0008153353,0.001424244,0.001623619,0.002533118,0.00160078,0.003182626,0.00188933],"category_scores_gemma":[0.03854831,0.000817819,0.0009088204,0.0008988879,0.004567063,0.005846406,0.004966218,0.002814424,0.001454189],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002110364,"about_ca_system_score_gemma":0.001326813,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004604654,"about_ca_topic_score_gemma":0.004044277,"domain_scores_codex":[0.983353,0.009695238,0.0008094554,0.002569455,0.002973327,0.0005995511],"domain_scores_gemma":[0.9665014,0.02068041,0.003923182,0.0068199,0.001498701,0.0005763617],"domain_codex":null,"domain_gemma":"reproducibility","domain_candidate":"reproducibility","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.008835626,0.001023801,0.09826311,0.001706413,0.0008309347,0.02526421,0.05322006,0.1663173,0.1141345,0.07910954,0.06216812,0.3891265],"study_design_scores_gemma":[0.0002025574,0.0009801731,0.01762227,0.0003687116,0.0002134989,0.008027966,0.003729835,0.8480942,0.0582798,0.02258322,0.03959015,0.0003076244],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7580932,0.001024136,0.2088336,0.009182459,0.0004514953,0.0009145216,0.001548467,0.008019745,0.01193245],"genre_scores_gemma":[0.923695,0.0001835785,0.06918599,0.001404688,0.0001247475,0.0001429915,0.0008249267,0.0003728747,0.004065143],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9912264,"threshold_uncertainty_score":0.04639983,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03774819822446671,"score_gpt":0.4296206950065799,"score_spread":0.3918724967821132,"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."}}