{"id":"W4404867262","doi":"10.1038/s41598-024-81170-y","title":"De-identification is not enough: a comparison between de-identified and synthetic clinical notes","year":2024,"lang":"en","type":"article","venue":"Scientific Reports","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":24,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Manitoba","funders":"National Human Genome Research Institute; U.S. National Library of Medicine; National Institute on Aging; Natural Sciences and Engineering Research Council of Canada; University of Texas Health Science Center at Houston; University of Manitoba; National Cancer Institute; National Science Foundation; National Institutes of Health; Cancer Prevention and Research Institute of Texas","keywords":"Computer science; Identification (biology); Inference; Synthetic data; Task (project management); Machine learning; Artificial intelligence; Generative model; Generative grammar; Safeguarding; Data mining; Data science; Medicine","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.008980951,0.0002255428,0.0003646599,0.0004192068,0.0004311091,0.004459448,0.006949849,0.0002545277,0.00003094945],"category_scores_gemma":[0.02011484,0.0002190978,0.000141126,0.001127488,0.0007116653,0.0009877569,0.01711795,0.0004827513,0.0001288635],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001350549,"about_ca_system_score_gemma":0.0003564702,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005790832,"about_ca_topic_score_gemma":0.000008050081,"domain_scores_codex":[0.995127,0.000198599,0.001267614,0.002051177,0.0007452469,0.000610359],"domain_scores_gemma":[0.9887825,0.0008716985,0.0003728365,0.009616033,0.0001440177,0.0002128811],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.000004837159,0.0001545543,0.06919115,0.0002432241,0.0001291113,0.0008363099,0.001401432,0.000008235042,0.01551005,0.002096195,0.5258315,0.3845934],"study_design_scores_gemma":[0.0001267135,0.00005625309,0.03412687,0.0003467369,0.0001147275,0.0006367759,0.00006756265,0.1858507,0.144667,0.5559468,0.07740782,0.0006520828],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5478981,0.001080936,0.414028,0.02475521,0.009691675,0.0004272744,0.00002102518,0.001859496,0.0002383133],"genre_scores_gemma":[0.945589,0.00003082022,0.05357609,0.00007293395,0.0001036852,0.00002766162,0.00002702355,0.00002045702,0.0005523078],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.5538507,"threshold_uncertainty_score":0.998423,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0863253993477095,"score_gpt":0.373935892853746,"score_spread":0.2876104935060365,"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."}}