China
Africa
Explore chapters and articles related to this topic
Security and Privacy Aspects of AI, IoT, Big Data and Blockchain in Healthcare Industry
Published in Pushpa Singh, Divya Mishra, Kirti Seth, Transformation in Healthcare with Emerging Technologies, 2022
Apoorva Joshi, Ambrish Kumar Sharma, Sanjeev Gour, Pratima Guatam
Methods of preventing the use of sensitive data of a person by hiding any data that may able to identify the patient information, either by removal of unique patient identifiers, or by the statistical method, require the patient to check or verify that information and that enough identifiers have been distorted. However, outsiders could be able to obtain external data by the de-identification technique in big data. A consequence is that de-identification is insufficient to secure the privacy of big data. Building powerful privacy-preserving methods to avoid the possibility of re-identification makes the technique more feasible. To improve this conventional methodology, we use the following methods:k-anonymityl-diversityt-closeness
An unsupervised embedding harmonization system for privacy-preserving data mining in healthcare
Published in IISE Transactions on Healthcare Systems Engineering, 2023
Mai Li, Ying Lin, Hua Chen, Rajender R. Aparasu
In this section, we review the privacy-preserving methods designed for EHR data. One of the traditional privacy-preserving methods is de-identification (Kushida et al., 2012). The de-identification method aims to remove the direct identifiers in medical records that can be used to identify the patient. According to HIPAA, there are 18 direct identifiers that are typically present in patients’ medical records, including names, geographic subdivisions smaller than a state (e.g., street address, city, and postcode), telephone numbers, full-face photographic images and so on. (HHSgov, 2012). De-identification helps share the data across multiple institutions or hospitals for further research while protecting patient privacy. However, limitations of many current de-identification systems include the inability to detect misspellings and proper names that share characteristics with non-protected health information (PHI) (e.g., the family name “Black”); restrictions on handling only certain types of data, such as discharge summaries; algorithms that are not designed to handle diverse PHI (e.g., hard-coded or embedded PHI in device-generated output files) and so on. Another challenge is to balance the degree of de-identification with operational cost, time, and labor.
A review of Automatic end-to-end De-Identification: Is High Accuracy the Only Metric?
Published in Applied Artificial Intelligence, 2020
Vithya Yogarajan, Bernhard Pfahringer, Michael Mayo
A superior de-identification system will not only meet legal requirements but will also help build societal consent by assuring the public that their privacy and medical data will be protected. This consent is vital if large-scale research involving medical records is to be accepted in the same way as, for example, Statistics New Zealand’s Integrated Data Infrastructure. Acceptance of the latter is arguably in part due to measures were taken by Statistics New Zealand to de-identify data (Ragupathy and Yogarajan (2018); Statistics New Zealand (2016)).