De-identifying Spanish medical texts - named entity recognition applied to radiology reports

Journal of Biomedical Semantics

Irene Pérez-Díez¹  Adolfo López-Cerdán¹  Raúl Pérez-Moraga²  María de la Iglesia-Vayá³  Jose-Maria Salinas-Serrano⁴ 

[1] FISABIO-CIPF Joint Research Unit in Biomedical Imaging. Fundació per al Foment de la Investigació Sanitària i Biomèdica (FISABIO), Av. de Catalunya 21, 46020, València, Spain;Bioinformatics and Biostatistics Unit. Centro de Investigación Príncipe Felipe (CIPF), Carrer d’Eduardo Primo Yúfera 3, 46012, València, Spain;FISABIO-CIPF Joint Research Unit in Biomedical Imaging. Fundació per al Foment de la Investigació Sanitària i Biomèdica (FISABIO), Av. de Catalunya 21, 46020, València, Spain;ESI International Chair@CEU-UCH, Departamento de Matemáticas, Física y Ciencias Tecnológicas, Universidad Cardenal Herrera-CEU, CEU Universities, Calle San Bartolomé 55, 46115, Alfafara del Patriarca, Spain;FISABIO-CIPF Joint Research Unit in Biomedical Imaging. Fundació per al Foment de la Investigació Sanitària i Biomèdica (FISABIO), Av. de Catalunya 21, 46020, València, Spain;Regional ministry of Universal Health and Public Health in Valencia, Carrer de Misser Mascó 31, 46010, València, Spain;CIBERSAM, ISCIII, Av. Blasco Ibáñez 15, 46010, València, Spain;Health Informatics Department, Hospital San Juan de Alicante, 03550, Sant Joan d’Alacant, Spain;

关键词: Natural language processing;    Named entity recognition;    Radiology reports;    Medical texts;    Spanish;   

DOI  :  10.1186/s13326-021-00236-2

来源: Springer

【 摘 要 】

BackgroundMedical texts such as radiology reports or electronic health records are a powerful source of data for researchers. Anonymization methods must be developed to de-identify documents containing personal information from both patients and medical staff. Although currently there are several anonymization strategies for the English language, they are also language-dependent. Here, we introduce a named entity recognition strategy for Spanish medical texts, translatable to other languages.ResultsWe tested 4 neural networks on our radiology reports dataset, achieving a recall of 97.18% of the identifying entities. Alongside, we developed a randomization algorithm to substitute the detected entities with new ones from the same category, making it virtually impossible to differentiate real data from synthetic data. The three best architectures were tested with the MEDDOCAN challenge dataset of electronic health records as an external test, achieving a recall of 69.18%.ConclusionsThe strategy proposed, combining named entity recognition tasks with randomization of entities, is suitable for Spanish radiology reports. It does not require a big training corpus, thus it could be easily extended to other languages and medical texts, such as electronic health records.

【 授权许可】

CC BY