Search

The growing use of artificial neural network (ANN) tools for computed tomography angiography (CTA) data analysis underscores the necessity for elevated data protection measures. We aimed to establish an automated defacing pipeline for CTA data. In this retrospective study, CTA data from multi-institutional cohorts were utilized to annotate facemasks (n = 100) and train an ANN model, subsequently tested on an external institution's dataset (n = 50) and compared to a publicly available defacing algorithm. Face detection (MTCNN) and verification (FaceNet) networks were applied to measure the similarity between the original and defaced CTA images. Dice similarity coefficient (DSC), face detection probability, and face similarity measures were calculated to evaluate model performance. The CTA-DEFACE model effectively segmented soft face tissue in CTA data achieving a DSC of 0.94 ± 0.02 (mean ± standard deviation) on the test set. Our model was benchmarked against a publicly available defacing algorithm. After applying face detection and verification networks, our model showed substantially reduced face detection probability (p < 0.001) and similarity to the original CTA image (p < 0.001). The CTA-DEFACE model enabled robust and precise defacing of CTA data. The trained network is publicly accessible at www.github.com/neuroAI-HD/CTA-DEFACE. Relevance statement: The ANN model CTA-DEFACE, developed for automatic defacing of CT angiography images, achieves significantly lower face detection probabilities and greater dissimilarity from the original images compared to a publicly available model. The algorithm has been externally validated and is publicly accessible. Key Points: The developed ANN model (CTA-DEFACE) automatically generates facemasks for CT angiography images. CTA-DEFACE offers superior deidentification capabilities compared to a publicly available model. By means of graphics processing unit optimization, our model ensures rapid processing of medical images. Our model underwent external validation, underscoring its reliability for real-world application. [ABSTRACT FROM AUTHOR]

Abstract The growing use of artificial neural network (ANN) tools for computed tomography angiography (CTA) data analysis underscores the necessity for elevated data protection measures. We aimed to establish an automated defacing pipeline for CTA data. In this retrospective study, CTA data from multi-institutional cohorts were utilized to annotate facemasks (n = 100) and train an ANN model, subsequently tested on an external institution’s dataset (n = 50) and compared to a publicly available defacing algorithm. Face detection (MTCNN) and verification (FaceNet) networks were applied to measure the similarity between the original and defaced CTA images. Dice similarity coefficient (DSC), face detection probability, and face similarity measures were calculated to evaluate model performance. The CTA-DEFACE model effectively segmented soft face tissue in CTA data achieving a DSC of 0.94 ± 0.02 (mean ± standard deviation) on the test set. Our model was benchmarked against a publicly available defacing algorithm. After applying face detection and verification networks, our model showed substantially reduced face detection probability (p

This paper presents a novel Anonymization Strategy Selection Framework that combines relaxed functional dependencies (RFDs) and particle swarm optimization (PSO) to balance data privacy and utility. Our approach extracts RFDs from datasets, generates diverse anonymization strategies using domain generalization hierarchies, and employs PSO for strategy optimization. We introduce a fitness function that balances k-anonymity and information loss. The framework's innovation lies in using RFDs to capture fine-grained data dependencies, enabling more nuanced anonymization. Evaluation on widely used UCI machine learning repository datasets show our framework outperforms existing techniques, achieving higher k-anonymity levels with lower information loss. Our adaptive approach generates hybrid strategies combining elements from multiple RFDs, resulting in superior privacy-utility trade-offs. This research advances privacy-preserving data publishing by providing a flexible, effective tool for generating anonymized datasets that maintain high utility for downstream analysis. [ABSTRACT FROM AUTHOR]

Healthcare institutions generate massive amounts of valuable patient data in the digital age. Finding the right balance between patient privacy and the demand for data-driven medical enhancements is essential. Since data privacy has become increasingly important, robust technologies must be developed to safeguard private data and allow meaningful exploration. This issue was addressed by ShinyAnonymizer, which was first created to anonymize health data. It achieves this by rendering anonymization methods easily available to users. The enhanced version of ShinyAnonymizer, with an essential improvement in performance, is presented in this study. We explain the merging of data analysis, visualization, and privacy-focused statistics paradigms with data anonymization, hashing, and encryption, offering researchers and data analysts an extensive collection of tools for trustworthy data management. [ABSTRACT FROM AUTHOR]

The emergence of data controllers as a novel term within data privacy laws, such as the General Data Protection Regulation (GDPR), has ushered in significant responsibilities. Stricter regulations prohibit the intentional sharing of personal records on the Internet. This research focuses on safeguarding data privacy, specifically in ubiquitous tabular formats across numerous websites. A novel approach employing a cell-key perturbation method is proposed, demonstrating efficacy in tabular formats. Addressing this challenge, we introduce the in-memory perturbation optimizer (IMPO) algorithm as a novel solution. The primary objective is to create and develop a platform that secures all personal data through a dispenser server, operating in near real-time. Also, it emphasizes the importance of balancing data utility with privacy protection to maintain the integrity and quality of the dataset. Experimental results reveal that the IMPO algorithm outperforms in terms of data accuracy. Additionally, the algorithm introduces an average time delay of 2 seconds, ensuring optimal time service for real-time datasets. [ABSTRACT FROM AUTHOR]

Edge storage is driven by the emerging edge computing and application intelligence, which makes the data anonymization become essential to guarantee the security of the data. The risk with data anonymization is that it can be re-identified, where anonoymized data is matched with available information to discover the individual. Appropriate data anonymization techniques can help on reducing the risk of re-identification. It is very challenging to protect data privacy against re-identification attacks without significantly reduce usefulness of the data. This work proposes an evaluation framework for data anonymization techniques to measure the risks of re-identification and usefulness of anonymized data. [ABSTRACT FROM AUTHOR]

This article introduces an advanced solution for anonymizing large-scale sensitive data, addressing the limitations of traditional approaches when applied to vast datasets. By leveraging the Spark distributed computing framework, we propose a method that parallelizes the data anonymization process, enhancing efficiency and scalability. Utilizing Spark's resilient distributed datasets (RDD), the approach integrates two primary operations, Map_RDD and ReduceByKey_RDD, to execute the anonymization tasks. Our comprehensive experimental evaluation demonstrates our solution's effectiveness and improved performance in preserving data privacy while balancing data utility and confidentiality. A significant contribution of our study is the development of a wide array of solutions for data owners, particularly notable for a 500 MB dataset at an anonymity level of K=100, where our methodology produces 832 unique solutions. This study also opens avenues for future research in applying different privacy models within the Spark ecosystem, such as l-diversity and t-closeness. [ABSTRACT FROM AUTHOR]

Abstract Artificial intelligence (AI) is revolutionizing the field of medical imaging, holding the potential to shift medicine from a reactive “sick-care” approach to a proactive focus on healthcare and prevention. The successful development of AI in this domain relies on access to large, comprehensive, and standardized real-world datasets that accurately represent diverse populations and diseases. However, images and data are sensitive, and as such, before using them in any way the data needs to be modified to protect the privacy of the patients. This paper explores the approaches in the domain of five EU projects working on the creation of ethically compliant and GDPR-regulated European medical imaging platforms, focused on cancer-related data. It presents the individual approaches to the de-identification of imaging data, and describes the problems and the solutions adopted in each case. Further, lessons learned are provided, enabling future projects to optimally handle the problem of data de-identification. Critical relevance statement This paper presents key approaches from five flagship EU projects for the de-identification of imaging and clinical data offering valuable insights and guidelines in the domain. Key Points ΑΙ models for health imaging require access to large amounts of data. Access to large imaging datasets requires an appropriate de-identification process. This paper provides de-identification guidelines from the AI for health imaging (AI4HI) projects.

Artificial intelligence (AI) is revolutionizing the field of medical imaging, holding the potential to shift medicine from a reactive "sick-care" approach to a proactive focus on healthcare and prevention. The successful development of AI in this domain relies on access to large, comprehensive, and standardized real-world datasets that accurately represent diverse populations and diseases. However, images and data are sensitive, and as such, before using them in any way the data needs to be modified to protect the privacy of the patients. This paper explores the approaches in the domain of five EU projects working on the creation of ethically compliant and GDPR-regulated European medical imaging platforms, focused on cancer-related data. It presents the individual approaches to the de-identification of imaging data, and describes the problems and the solutions adopted in each case. Further, lessons learned are provided, enabling future projects to optimally handle the problem of data de-identification. Critical relevance statement: This paper presents key approaches from five flagship EU projects for the de-identification of imaging and clinical data offering valuable insights and guidelines in the domain. Key Points: ΑΙ models for health imaging require access to large amounts of data. Access to large imaging datasets requires an appropriate de-identification process. This paper provides de-identification guidelines from the AI for health imaging (AI4HI) projects. [ABSTRACT FROM AUTHOR]

The contemporary landscape of electricity marketing data utilization is characterized by increased openness, heightened data circulation, and more intricate interaction contexts. Throughout the entire lifecycle of data, the persistent threat of leakage is ever-present. In this study, we introduce a novel electricity data anonymization model, termed EPA-GAN, which relies on table generation. In comparison to existing methodologies, our model extends the foundation of generative adversarial networks by incorporating feature encoders and feedback mechanisms. This adaptation enables the generation of anonymized data with heightened practicality and similarity to the original data, specifically tailored for mixed data types, thereby achieving a deliberate decoupling from the source data. Our proposed approach initiates by parsing the original JSON file, encoding it based on variable types and features using distinct feature encoders. Subsequently, a generative adversarial network, enhanced with information, downstream, generator losses, and the Was + GP modification, is employed to generate anonymized data. The introduction of random noise fortifies privacy protection during the data generation process. Experimental validation attests to a conspicuous reduction in both machine learning utility and statistical dissimilarity between the data synthesized by our proposed anonymization model and the original dataset. This substantiates the model's efficacy in replacing the original data for mining analysis and data sharing, thereby effectively safeguarding the privacy of the source data. [ABSTRACT FROM AUTHOR]

With the vigorously developed services of cloud computing, it is relatively easier and more convenient for organizations or enterprises to open data on clouds. However, as personal information in electronic data becomes more massive and detailed, how to balance data opening and personal privacy has become a critical issue. In this paper, we propose the Theseus Data Synthesis Approach (TDSA), which generates synthetic data by replacing partial records until no record from the original dataset remains. Unlike other data anonymization works such as k-anonymity and differential privacy, which encountered limitations and challenges when applying to real-world scenarios. In our work, Since there are no real data, personal privacy is definitely preserved. We also analyze the quality and utility of the synthetic dataset and make comparisons with related works. We conclude that with our scheme, opening useful data on clouds and preserving personal privacy can be simultaneously achieved.

The exponential growth of data in the information age poses several threats to the privacy and safety of digital service users. Existing legislation, such as the GDPR in Europe and the CCPA in California, defines frameworks and guidelines to promote personal privacy but leaves freedom in the choice of means to achieve privacy. Data anonymization techniques remove information that can be used to identify individuals from the dataset, either through suppression, generalization, anatomization, permutation, or perturbation. Information suppression remains the most common, safe, and widely applicable anonymization method, though at a high data utility cost. In this paper, we argue that even information suppression may not be sufficient in some cases. We study the case of a dataset that describes the shopping habits of a grocery store’s customers. All identifiers and quasi-identifiers are removed from the dataset by suppression. However, by augmenting the data in a novel multi-step, iterative process, and building a neural network enriched with prior knowledge, we show that most sensitive information can be retrieved with an accuracy of 80%.

Data anonymization is a technique that safeguards individuals' privacy by modifying attribute values in published data. However, increased modifications enhance privacy but diminish the utility of published data, necessitating a balance between privacy and utility levels. K-Anonymity is a crucial anonymization technique that generates k-anonymous clusters, where the probability of disclosing a record is 1 / k . However, k-anonymity fails to protect against attribute disclosure when the diversity of sensitive values within the anonymous cluster is insufficient. Several techniques have been proposed to address this issue, among which t-closeness is considered one of the most robust privacy techniques. In this paper, we propose a novel approach employing a greedy and information-theoretic clustering-based algorithm to achieve strict privacy protection. The proposed anonymization algorithm commences by clustering the data based on both the similarity of quasi-identifier values and the diversity of sensitive attribute values. In the subsequent adjustment phase, the algorithm splits and merges the clusters to ensure that they each possess at least k members and adhere to the t-closeness requirements. Finally, the algorithm replaces the quasi-identifier values of the records in each cluster with the values of the cluster center to attain k-anonymity and t-closeness. Experimental results on three microdata sets from Facebook, Twitter, and Google+ demonstrate the proposed algorithm's ability to preserve the utility of released data by minimizing the modifications of attribute values while satisfying the k-anonymity and t-closeness constraints. [ABSTRACT FROM AUTHOR]

Objectives Public healthcare data have become crucial to the advancement of medicine, and recent changes in legal structure on privacy protection have expanded access to these data with pseudonymization. Recent debates on public healthcare data use by private insurance companies have shown large discrepancies in perceptions among the general public, healthcare professionals, private companies, and lawmakers. This study examined public attitudes toward the secondary use of public data, focusing on differences between public and private entities. Methods An online survey was conducted from January 11 to 24, 2022, involving a random sample of adults between 19 and 65 of age in 17 provinces, guided by the August 2021 census. Results The final survey analysis included 1,370 participants. Most participants were aware of health data collection (72.5%) and recent changes in legal structures (61.4%) but were reluctant to share their pseudonymized raw data (51.8%). Overall, they were favorable toward data use by public agencies but disfavored use by private entities, notably marketing and private insurance companies. Concerns were frequently noted regarding commercial use of data and data breaches. Among the respondents, 50.9% were negative about the use of public healthcare data by private insurance companies, 22.9% favored this use, and 1.9% were “very positive.” Conclusions This survey revealed a low understanding among key stakeholders regarding digital health data use, which is hindering the realization of the full potential of public healthcare data. This survey provides a basis for future policy developments and advocacy for the secondary use of health data.

Abstract Cryptocurrencies, particularly Bitcoin, have garnered attention for their potential in anonymous transactions. However, their anonymity has often been compromised by deanonymization attacks. To counter this, mixing services have been introduced. While they enhance privacy, they obscure fund traceability. This study seeks to demystify transactions linked to these services, shedding light on pathways of concealed and laundered money. We propose a method to identify and classify transactions and addresses of major mixing services in Bitcoin. Unlike previous research focusing on older techniques like CoinJoin, we emphasize modern mixing services. We gathered labelled data by transacting with three prominent mixers (MixTum, Blemder, and CryptoMixer) and identified recurring patterns. Using these patterns, an algorithm was created to pinpoint mixing transactions and distinguish mixer‐related addresses. The algorithm achieved a remarkable recall rate of 100%. Given the lack of clear ground truth and the vast number of unlabelled transactions, ensuring accuracy was a challenge. However, by analyzing a set of non‐mixing transactions with our model, it was confirmed that the high recall rate was not misleading. This work provides a significant advancement in monitoring mixing transactions, presenting a valuable tool against fraud and money laundering in cryptocurrency networks.

Healthcare institutions generate massive amounts of valuable patient data in the digital age. Finding the right balance between patient privacy and the demand for data-driven medical enhancements is essential. Since data privacy has become increasingly important, robust technologies must be developed to safeguard private data and allow meaningful exploration. This issue was addressed by ShinyAnonymizer, which was first created to anonymize health data. It achieves this by rendering anonymization methods easily available to users. The enhanced version of ShinyAnonymizer, with an essential improvement in performance, is presented in this study. We explain the merging of data analysis, visualization, and privacy-focused statistics paradigms with data anonymization, hashing, and encryption, offering researchers and data analysts an extensive collection of tools for trustworthy data management.

Privacy protection is absolutely imperative for data releases when the utilization of public data and big data is getting popular. In this paper, data anonymization methods using rough set-based rule induction are investigated. It has been shown that many rules with imprecise conclusions can improve the classification accuracy of the rule-based classifier. Data anonymization methods utilizing rules with imprecise conclusions are proposed. The data tables anonymized by one of the proposed methods can preserve the classification accuracy of the rules induced from them. The proposed methods as well as conventional data anonymization methods are compared from two viewpoints: the classification accuracy of rules induced from the anonymized data table and the preservation of data anonymity. The results show the usefulness of the proposed methods. [ABSTRACT FROM AUTHOR]

The healthcare field remains one of the most important social and economic challenges worldwide, demanding new and more advanced solutions from science and technology. The developments in Industry 4.0 have brought tremendous improvements in the healthcare industry such as better quality of treatment, improved communication, remote monitoring, and lower cost. Therefore, sharing health data with healthcare providers is important for the continuation of such improvements. In general, health data contain sensitive information about individuals. Therefore, in accordance with privacy regulations and ethical requirements, it is essential to protect patients' privacy before sharing data for medical research. In this study, a new dataset is created with the data collected from e-Health platforms where patients and doctors meet. Some k-anonymization-based algorithms, which are frequently used in the literature, have been applied to this created health dataset. In order to reveal the success of the algorithms applied to the health dataset, various information metrics have been used and the results obtained from the experiments are presented in detail. [ABSTRACT FROM AUTHOR]

Developing a privacy-preserving data publishing algorithm that stops individuals from disclosing their identities while not ignoring data utility remains an important goal to achieve. Because finding the trade-off between data privacy and data utility is an NP-hard problem and also a current research area. When existing approaches are investigated, one of the most significant difficulties discovered is the presence of outlier data in the datasets. Outlier data has a negative impact on data utility. Furthermore, k-anonymity algorithms, which are commonly used in the literature, do not provide adequate protection against outlier data. In this study, a new data anonymization algorithm is devised and tested for boosting data utility by incorporating an outlier data detection mechanism into the Mondrian algorithm. The connectivity-based outlier factor (COF) algorithm is used to detect outliers. Mondrian is selected because of its capacity to anonymize multidimensional data while meeting the needs of real-world data. COF, on the other hand, is used to discover outliers in high-dimensional datasets with complicated structures. The proposed algorithm generates more equivalence classes than the Mondrian algorithm and provides greater data utility than previous algorithms based on k-anonymization. In addition, it outperforms other algorithms in the discernibility metric (DM), normalized average equivalence class size (Cavg), global certainty penalty (GCP), query error rate, classification accuracy (CA), and F-measure metrics. Moreover, the increase in the values of the GCP and error rate metrics demonstrates that the proposed algorithm facilitates obtaining higher data utility by grouping closer data points when compared to other algorithms. [ABSTRACT FROM AUTHOR]

Abstract Artificial intelligence (AI) is transforming the field of medical imaging and has the potential to bring medicine from the era of ‘sick-care’ to the era of healthcare and prevention. The development of AI requires access to large, complete, and harmonized real-world datasets, representative of the population, and disease diversity. However, to date, efforts are fragmented, based on single–institution, size-limited, and annotation-limited datasets. Available public datasets (e.g., The Cancer Imaging Archive, TCIA, USA) are limited in scope, making model generalizability really difficult. In this direction, five European Union projects are currently working on the development of big data infrastructures that will enable European, ethically and General Data Protection Regulation-compliant, quality-controlled, cancer-related, medical imaging platforms, in which both large-scale data and AI algorithms will coexist. The vision is to create sustainable AI cloud-based platforms for the development, implementation, verification, and validation of trustable, usable, and reliable AI models for addressing specific unmet needs regarding cancer care provision. In this paper, we present an overview of the development efforts highlighting challenges and approaches selected providing valuable feedback to future attempts in the area. Key points • Artificial intelligence models for health imaging require access to large amounts of harmonized imaging data and metadata. • Main infrastructures adopted either collect centrally anonymized data or enable access to pseudonymized distributed data. • Developing a common data model for storing all relevant information is a challenge. • Trust of data providers in data sharing initiatives is essential. • An online European Union meta-tool-repository is a necessity minimizing effort duplication for the various projects in the area.

The increasing need for data privacy and the rising complexity of data environments necessitate robust data anonymization techniques to safeguard personal and sensitive information. A multi-model approach to data anonymization can strike an optimal balance between privacy protection and data utility, integrating techniques such as data masking, differential privacy, machine learning algorithms, blockchain technology, and data encryption. This article introduces a Security-Centric Enterprise Data Anonymization Governance Model, a structured framework for managing data privacy across healthcare, finance, and government industries. The model ensures adherence to best practices and compliance with legal and regulatory requirements. The article addresses challenges in implementing data anonymization techniques, including maintaining data utility and preventing reidentification, by advocating for a multi-model approach that combines various technologies and methods. We suggest that by adopting this holistic approach, organizations can enhance their data protection measures and foster a culture of data privacy. [ABSTRACT FROM AUTHOR]

Artificial intelligence (AI) is transforming the field of medical imaging and has the potential to bring medicine from the era of 'sick-care' to the era of healthcare and prevention. The development of AI requires access to large, complete, and harmonized real-world datasets, representative of the population, and disease diversity. However, to date, efforts are fragmented, based on single–institution, size-limited, and annotation-limited datasets. Available public datasets (e.g., The Cancer Imaging Archive, TCIA, USA) are limited in scope, making model generalizability really difficult. In this direction, five European Union projects are currently working on the development of big data infrastructures that will enable European, ethically and General Data Protection Regulation-compliant, quality-controlled, cancer-related, medical imaging platforms, in which both large-scale data and AI algorithms will coexist. The vision is to create sustainable AI cloud-based platforms for the development, implementation, verification, and validation of trustable, usable, and reliable AI models for addressing specific unmet needs regarding cancer care provision. In this paper, we present an overview of the development efforts highlighting challenges and approaches selected providing valuable feedback to future attempts in the area. Key points • Artificial intelligence models for health imaging require access to large amounts of harmonized imaging data and metadata. • Main infrastructures adopted either collect centrally anonymized data or enable access to pseudonymized distributed data. • Developing a common data model for storing all relevant information is a challenge. • Trust of data providers in data sharing initiatives is essential. • An online European Union meta-tool-repository is a necessity minimizing effort duplication for the various projects in the area. [ABSTRACT FROM AUTHOR]

The Internet of Things (IoT) has shown rapid growth in recent years. However, it presents challenges related to the lack of standardization of communication produced by different types of devices. Another problem area is the security and privacy of data generated by IoT devices. Thus, with the focus on grouping, analyzing, and classifying existing data security and privacy methods in IoT, based on data anonymization, we have conducted a Systematic Literature Review (SLR). We have therefore reviewed the history of works developing solutions for security and privacy in the IoT, particularly data anonymization and the leading technologies used by researchers in their work. We also discussed the challenges and future directions for research. The objective of the work is to give order to the main approaches that promise to provide or facilitate data privacy using anonymization in the IoT area. The study's results can help us understand the best anonymization techniques to provide data security and privacy in IoT environments. In addition, the findings can also help us understand the limitations of existing approaches and identify areas for improvement. The results found in most of the studies analyzed indicate a lack of consensus in the following areas: (i) with regard to a solution with a standardized methodology to be applied in all scenarios that encompass IoT; (ii) the use of different techniques to anonymize the data; and (iii), the resolution of privacy issues. On the other hand, results made available by the k-anonymity technique proved efficient in combination with other techniques. In this context, data privacy presents one of the main challenges for broadening secure domains in applying privacy with anonymity. [ABSTRACT FROM AUTHOR]

Data anonymization is a technique that safeguards individuals’ privacy by modifying attribute values in published data. However, increased modifications enhance privacy but diminish the utility of published data, necessitating a balance between privacy and utility levels. K-Anonymity is a crucial anonymization technique that generates k-anonymous clusters, where the probability of disclosing a record is 1/k. However, k-anonymity fails to protect against attribute disclosure when the diversity of sensitive values within the anonymous cluster is insufficient. Several techniques have been proposed to address this issue, among which t-closeness is considered one of the most robust privacy techniques. In this paper, we propose a novel approach employing a greedy and information-theoretic clustering-based algorithm to achieve strict privacy protection. The proposed anonymization algorithm commences by clustering the data based on both the similarity of quasi-identifier values and the diversity of sensitive attribute values. In the subsequent adjustment phase, the algorithm splits and merges the clusters to ensure that they each possess at least k members and adhere to the t-closeness requirements. Finally, the algorithm replaces the quasi-identifier values of the records in each cluster with the values of the cluster center to attain k-anonymity and t-closeness. Experimental results on three microdata sets from Facebook, Twitter, and Google+ demonstrate the proposed algorithm’s ability to preserve the utility of released data by minimizing the modifications of attribute values while satisfying the k-anonymity and t-closeness constraints.

The article presents an approach to data anonymization with the use of generally available tools. The focus is put on the practical aspects of using open-source tools in conjunction with programming libraries provided by suppliers of industrial control systems. This universal approach shows the possibilities of using various operating systems as a platform for process data anonymization. An additional advantage of the described approach is the ease of integration with various types of advanced data analysis tools based both on the out-of-the-box approach (e.g., business intelligence tools) as well as customized solutions. The discussed case describes the anonymization of data for the needs of sensitive analysis by a wider group of recipients during the construction of a predictive model used to support decisions. [ABSTRACT FROM AUTHOR]

Clinical data analysis could lead to breakthroughs. However, clinical data contain sensitive information about participants that could be utilized for unethical activities, such as blackmailing, identity theft, mass surveillance, or social engineering. Data anonymization is a standard step during data collection, before sharing, to overcome the risk of disclosure. However, conventional data anonymization techniques are not foolproof and also hinder the opportunity for personalized evaluations. Much research has been done for synthetic data generation using generative adversarial networks and many other machine learning methods; however, these methods are either not free to use or are limited in capacity. This study evaluates the performance of an emerging tool named synthpop, an R package producing synthetic data as an alternative approach for data anonymization. This paper establishes data standards derived from the original data set based on the utilities and quality of information and measures variations in the synthetic data set to evaluate the performance of the data synthesis process. The methods to assess the utility of the synthetic data set can be broadly divided into two approaches: general utility and specific utility. General utility assesses whether synthetic data have overall similarities in the statistical properties and multivariate relationships with the original data set. Simultaneously, the specific utility assesses the similarity of a fitted model's performance on the synthetic data to its performance on the original data. The quality of information is assessed by comparing variations in entropy bits and mutual information to response variables within the original and synthetic data sets. The study reveals that synthetic data succeeded at all utility tests with a statistically non-significant difference and not only preserved the utilities but also preserved the complexity of the original data set according to the data standard established in this study. Therefore, synthpop fulfills all the necessities and unfolds a wide range of opportunities for the research community, including easy data sharing and information protection. [ABSTRACT FROM AUTHOR]

Objectives De-identifying protected health information (PHI) in medical documents is important, and a prerequisite to de-identification is the identification of PHI entity names in clinical documents. This study aimed to compare the performance of three pre-training models that have recently attracted significant attention and to determine which model is more suitable for PHI recognition. Methods We compared the PHI recognition performance of deep learning models using the i2b2 2014 dataset. We used the three pre-training models—namely, bidirectional encoder representations from transformers (BERT), robustly optimized BERT pre-training approach (RoBERTa), and XLNet (model built based on Transformer-XL)—to detect PHI. After the dataset was tokenized, it was processed using an inside-outside-beginning tagging scheme and WordPiece-tokenized to place it into these models. Further, the PHI recognition performance was investigated using BERT, RoBERTa, and XLNet. Results Comparing the PHI recognition performance of the three models, it was confirmed that XLNet had a superior F1-score of 96.29%. In addition, when checking PHI entity performance evaluation, RoBERTa and XLNet showed a 30% improvement in performance compared to BERT. Conclusions Among the pre-training models used in this study, XLNet exhibited superior performance because word embedding was well constructed using the two-stream self-attention method. In addition, compared to BERT, RoBERTa and XLNet showed superior performance, indicating that they were more effective in grasping the context.

Code-mixing is a phenomenon where at least two languages are combined in a hybrid manner in the context of a single conversation. The use of mixed language is widespread in multilingual and multicultural countries and poses significant challenges for the development of automated language processing tools. In Taiwan’s electronic health record (EHR) systems, unstructured EHR texts are usually represented in a mixture of English and Chinese which increases the difficulty for de-identification and synthetization of protected health information (PHI). We explored this problem by applying several state-of-the-art pre-trained mono- and multilingual language models and propose to exploit the principle-based approach (PBA) for the tasks of PHI recognition and resynthesis on a code-mixed EHR corpus annotated with 6 main categories and 25 subcategories of PHIs. A hierarchical principle slot schema is defined in the PBA to encode knowledge of code-mixed PHIs and utilize slots to learn from the training set to assemble principles for recognizing PHI mentions and synthesizing surrogates simultaneously. In addition, a semantic disambiguation process is implemented to disambiguate ambiguous PHI categories in the de-identification process and to dynamically extend the knowledge encoded in PBA during the knowledge augmentation process. The experiment results demonstrate that the proposed method can achieve the best micro- and macro-F-scores in comparison to the other mono- and multilingual language models fine-tuned on our code-mixed corpus.

With the emerging of e-commerce, package theft is at a high level: It is reported that 1.7 million packages are stolen or lost every day in the U.S. in 2020, which costs $25 million every day for the lost packages and the service. Information leakage during transportation is an important reason for theft since thieves can identify which truck is the target that contains the valuable products. In this paper, we address the privacy and security issues in bin-packing, which is an algorithm used in delivery centers to determine which packages should be loaded together to a certain truck. Data such as the weight of the packages is needed when assigning items into trucks, which can be called bins. However, the information is sensitive and can be used to identify the contents in the package. To provide security and privacy during bin-packing, we propose two different privacy-preserving data publishing methods. Both approaches use differential privacy (DP) to hide the existence of any specific package to prevent it from being identified by malicious users. The first approach combines differential privacy with k-anonymity, and the other one applies clustering before differential privacy. Our extensive analyses and experimental results clearly show that our proposed approaches have better privacy guarantees, better efficiency, and better performance than the existing works that use either differential privacy or k-anonymity.

Data sharing between different organizations is an essential process in today’s connected world. However, recently there were many concerns about data sharing as sharing sensitive information can jeopardize users’ privacy. To preserve privacy, organizations use anonymization techniques to conceal users’ sensitive data. However, these techniques are vulnerable to de-anonymization attacks which aim to identify individual records within a dataset. In this paper, a two-tier mathematical framework is proposed for analyzing and mitigating the de-anonymization attacks, by studying the interactions between sharing organizations, a data collector, and an attacker. In the first level, a game-theoretic model is proposed to enable sharing organizations to optimally select their anonymization levels for k-anonymization under two potential attacks: background-knowledge attack and homogeneity attack. In the second level, a contract-theoretic model is proposed to enable the data collector to optimally reward the organizations for their data. The formulated problems are studied under single-time sharing and repeated sharing scenarios. Different Nash equilibria for the proposed game and the optimal solution of the contract-based problem are analytically derived for both scenarios. Simulation results show that the organizations can optimally select their anonymization levels, while the data collector can benefit from incentivizing the organizations to share their data.

Precision medicine relies on molecular and systems biology methods as well as bidirectional association studies of phenotypes and (high-throughput) genomic data. However, the integrated use of such data often faces obstacles, especially in regards to data protection. An important prerequisite for research data processing is usually informed consent. But collecting consent is not always feasible, in particular when data are to be analyzed retrospectively. For phenotype data, anonymization, i.e. the altering of data in such a way that individuals cannot be identified, can provide an alternative. Several re-identification attacks have shown that this is a complex task and that simply removing directly identifying attributes such as names is usually not enough. More formal approaches are needed that use mathematical models to quantify risks and guide their reduction. Due to the complexity of these techniques, it is challenging and not advisable to implement them from scratch. Open software libraries and tools can provide a robust alternative. However, also the range of available anonymization tools is heterogeneous and obtaining an overview of their strengths and weaknesses is difficult due to the complexity of the problem space. We therefore performed a systematic review of open anonymization tools for structured phenotype data described in the literature between 1990 and 2021. Through a two-step eligibility assessment process, we selected 13 tools for an in-depth analysis. By comparing the supported anonymization techniques and further aspects, such as maturity, we derive recommendations for tools to use for anonymizing phenotype datasets with different properties. [ABSTRACT FROM AUTHOR]

To facilitate the secondary usage of electronic health record data for research, the University of California, San Francisco (UCSF) recently implemented a clinical data warehouse including, among other data, deidentified clinical notes and reports, which are available to UCSF researchers without Institutional Review Board approval. For deidentification of these notes, most of the Health Insurance Portability and Accountability Act identifiers are redacted, but dates are transformed by shifting all dates for a patient back by the same random number of days. We describe an issue in which nonspecific (ie, excess) transformation of nondate, date-like text by this deidentification process enables reidentification of all dates, including birthdates, for certain patients. This issue undercuts the common assumption that excess deidentification is a safe tradeoff to protect patient privacy. We present this issue as a caution to other institutions that may also be considering releasing deidentified notes for research. [ABSTRACT FROM AUTHOR]

Identifying protected health identifiers (PHIs) in free text is core to anonymising clinical notes for research. There is no standard metric for the quality of a PHI detection algorithm, resulting in papers reporting heterogeneous results. CensorCheck is presented, which computes specialised metrics for evaluating PHI detection taking into account the peculiarities of anonymisation tasks.