Membership Disclosure Research Articles

Cluster-based anonymity model and algorithm for 1:1 dataset with a single sensitive attribute using machine learning technique

Privacy is a significant issue that requires consideration in all applications. Data collected from various individuals and organizations must be disclosed to the public or private parties for analysis and research purposes. The collected data are studied and analyzed digitally for the extraction of various useful patterns for decision-making research purposes. Privacy-preserving data publishing is significant as privacy violations in the patient’s data may have an adverse effect on the individual positive reputation. An efficient Cluster Based anonymity model has been proposed to anonymizes the 1:1 dataset with a single sensitive attribute through the introduction of a concept named “Semi-sensitive attribute.” Based on correlation, the attributes are categorized as quasi-identifier and semi-sensitive attributes. The k-anonymity is implemented on the quasi-identifier with the semi-sensitive attribute table and Fuzzy c-means clustering has been implemented to fix a range of values for anonymizing the semi-sensitive attributes. The disease is considered a sensitive attribute as the research work focuses on the medical dataset. The proposed model is demonstrated to resist the three privacy attacks such as, i)Identity Disclosure, ii) Attribute Disclosure, and iii) Membership Disclosure. The utility loss is calculated for each row and utility loss of each record are aggregated and considered as the total information loss for each attribute. Cluster Based anonymity model measured the utility loss for all the attributes and the average utility loss for the anonymized patient dataset is 3.78%.

An evaluation of the replicability of analyses using synthetic health data

Synthetic data generation is being increasingly used as a privacy preserving approach for sharing health data. In addition to protecting privacy, it is important to ensure that generated data has high utility. A common way to assess utility is the ability of synthetic data to replicate results from the real data. Replicability has been defined using two criteria: (a) replicate the results of the analyses on real data, and (b) ensure valid population inferences from the synthetic data. A simulation study using three heterogeneous real-world datasets evaluated the replicability of logistic regression workloads. Eight replicability metrics were evaluated: decision agreement, estimate agreement, standardized difference, confidence interval overlap, bias, confidence interval coverage, statistical power, and precision (empirical SE). The analysis of synthetic data used a multiple imputation approach whereby up to 20 datasets were generated and the fitted logistic regression models were combined using combining rules for fully synthetic datasets. The effects of synthetic data amplification were evaluated, and two types of generative models were used: sequential synthesis using boosted decision trees and a generative adversarial network (GAN). Privacy risk was evaluated using a membership disclosure metric. For sequential synthesis, adjusted model parameters after combining at least ten synthetic datasets gave high decision and estimate agreement, low standardized difference, as well as high confidence interval overlap, low bias, the confidence interval had nominal coverage, and power close to the nominal level. Amplification had only a marginal benefit. Confidence interval coverage from a single synthetic dataset without applying combining rules were erroneous, and statistical power, as expected, was artificially inflated when amplification was used. Sequential synthesis performed considerably better than the GAN across multiple datasets. Membership disclosure risk was low for all datasets and models. For replicable results, the statistical analysis of fully synthetic data should be based on at least ten generated datasets of the same size as the original whose analyses results are combined. Analysis results from synthetic data without applying combining rules can be misleading. Replicability results are dependent on the type of generative model used, with our study suggesting that sequential synthesis has good replicability characteristics for common health research workloads.

Evaluating the Utility and Privacy of Synthetic Breast Cancer Clinical Trial Data Sets.

There is strong interest from patients, researchers, the pharmaceutical industry, medical journal editors, funders of research, and regulators in sharing clinical trial data for secondary analysis. However, data access remains a challenge because of concerns about patient privacy. It has been argued that synthetic data generation (SDG) is an effective way to address these privacy concerns. There is a dearth of evidence supporting this on oncology clinical trial data sets, and on the utility of privacy-preserving synthetic data. The objective of the proposed study is to validate the utility and privacy risks of synthetic clinical trial data sets across multiple SDG techniques. We synthesized data sets from eight breast cancer clinical trial data sets using three types of generative models: sequential synthesis, conditional generative adversarial network, and variational autoencoder. Synthetic data utility was evaluated by replicating the published analyses on the synthetic data and assessing concordance of effect estimates and CIs between real and synthetic data. Privacy was evaluated by measuring attribution disclosure risk and membership disclosure risk. Utility was highest using the sequential synthesis method where all results were replicable and the CI overlap most similar or higher for seven of eight data sets. Both types of privacy risks were low across all three types of generative models. Synthetic data using sequential synthesis methods can act as a proxy for real clinical trial data sets, and simultaneously have low privacy risks. This type of generative model can be one way to enable broader sharing of clinical trial data.

A comparison of synthetic data generation and federated analysis for enabling international evaluations of cardiovascular health

Sharing health data for research purposes across international jurisdictions has been a challenge due to privacy concerns. Two privacy enhancing technologies that can enable such sharing are synthetic data generation (SDG) and federated analysis, but their relative strengths and weaknesses have not been evaluated thus far. In this study we compared SDG with federated analysis to enable such international comparative studies. The objective of the analysis was to assess country-level differences in the role of sex on cardiovascular health (CVH) using a pooled dataset of Canadian and Austrian individuals. The Canadian data was synthesized and sent to the Austrian team for analysis. The utility of the pooled (synthetic Canadian + real Austrian) dataset was evaluated by comparing the regression results from the two approaches. The privacy of the Canadian synthetic data was assessed using a membership disclosure test which showed an F1 score of 0.001, indicating low privacy risk. The outcome variable of interest was CVH, calculated through a modified CANHEART index. The main and interaction effect parameter estimates of the federated and pooled analyses were consistent and directionally the same. It took approximately one month to set up the synthetic data generation platform and generate the synthetic data, whereas it took over 1.5 years to set up the federated analysis system. Synthetic data generation can be an efficient and effective tool for enabling multi-jurisdictional studies while addressing privacy concerns.

Can synthetic data accurately mimic oncology clinical trials?

1554 Background: There is strong interest by researchers, the pharmaceutical industry, medical journal editors, funders of research, and regulators in sharing clinical trial data. Reusing data extracts the most utility possible from patient contributions. The majority of patients do want to share their data for secondary research purposes. However, data access for secondary analysis remains a challenge. A key reason why individual-level data is not made directly available to data users by authors and data custodians is concern over breaches of patient privacy. Synthetic data generation (SDG) is an effective way to address privacy concerns that can enable the broader sharing of clinical trial datasets. However, a key question is whether the reproducibility of the generated data is adequate to draw reliable conclusions. Methods: We synthesized datasets from five pragmatic breast cancer clinical trials performed by the REaCT group (https://react.ohri.ca/). A sequential synthesis method, a type of machine learning was performed. The published analysis of each trial was repeated on each synthetic dataset to evaluate reproducibility. We evaluated reproducibility on three criteria: (a) decision agreement: the direction and statistical significance of the primary endpoint effect estimates are the same as the real data, (b) estimate agreement: the parameter estimates from the synthetic data are within the 95% confidence interval of the real data, and (c) the confidence interval overlap between real and synthetic parameters is above 50%. In addition, we evaluated privacy using a membership disclosure metric. This evaluates the ability of an adversary to determine that a target individual was in the original dataset using the synthetic data, computed as an F1 classification accuracy score. Results: Our results show that decision and estimate agreements held true across all five trials, and the confidence interval overlap was high. The risks of membership disclosure are all below the established 0.2 threshold. Conclusions: In this study, we were able to successfully generate synthetic datasets that accurately replicated original data from 5 oncology trials and yielded the same results as in the original published studies, with a very low risk of membership disclosure. With proper modeling techniques, synthetic datasets can play a key role in data democratization and the reuse of oncology clinical trials.[Table: see text]

AN EVALUATION OF DATA ANONYMIZATION METHODS FOR DATA PUBLISHING

Data anonymization approaches have been the focus of research recently for several types of structured data, including tabular, graph, and item set data. In this article, we provide a succinct yet thorough assessment of a number of anonymization approaches, including generalisation and bucketization, which have been created for publishing microdata while protecting privacy. Recent research has demonstrated that generalisation results in significant information loss, particularly for high-dimensional data. Bucketization, however, does not stop membership disclosure. While slicing both prevents membership disclosure and preserves the data’s superior utility than generalisation. The practical methods that can be employed to provide improved data utility and handle high-dimensional data are the main emphasis of this research.

Validating a membership disclosure metric for synthetic health data.

BackgroundOne of the increasingly accepted methods to evaluate the privacy of synthetic data is by measuring the risk of membership disclosure. This is a measure of the F1 accuracy that an adversary would correctly ascertain that a target individual from the same population as the real data is in the dataset used to train the generative model, and is commonly estimated using a data partitioning methodology with a 0.5 partitioning parameter.ObjectiveValidate the membership disclosure F1 score, evaluate and improve the parametrization of the partitioning method, and provide a benchmark for its interpretation.Materials and methodsWe performed a simulated membership disclosure attack on 4 population datasets: an Ontario COVID-19 dataset, a state hospital discharge dataset, a national health survey, and an international COVID-19 behavioral survey. Two generative methods were evaluated: sequential synthesis and a generative adversarial network. A theoretical analysis and a simulation were used to determine the correct partitioning parameter that would give the same F1 score as a ground truth simulated membership disclosure attack.ResultsThe default 0.5 parameter can give quite inaccurate membership disclosure values. The proportion of records from the training dataset in the attack dataset must be equal to the sampling fraction of the real dataset from the population. The approach is demonstrated on 7 clinical trial datasets.ConclusionsOur proposed parameterization, as well as interpretation and generative model training guidance provide a theoretically and empirically grounded basis for evaluating and managing membership disclosure risk for synthetic data.

Slicing-Based Enhanced Method for Privacy-Preserving in Publishing Big Data

Publishing big data and making it accessible to researchers is important for knowledge building as it helps in applying highly efficient methods to plan, conduct, and assess scientific research. However, publishing and processing big data poses a privacy concern related to protecting individuals’ sensitive information while maintaining the usability of the published data. Several anonymization methods, such as slicing and merging, have been designed as solutions to the privacy concerns for publishing big data. However, the major drawback of merging and slicing is the random permutation procedure, which does not always guarantee complete protection against attribute or membership disclosure. Moreover, merging procedures may generate many fake tuples, leading to a loss of data utility and subsequent erroneous knowledge extraction. This study therefore proposes a slicing-based enhanced method for privacy-preserving big data publishing while maintaining the data utility. In particular, the proposed method distributes the data into horizontal and vertical partitions. The lower and upper protection levels are then used to identify the unique and identical attributes’ values. The unique and identical attributes are swapped to ensure the published big data is protected from disclosure risks. The outcome of the experiments demonstrates that the proposed method could maintain data utility and provide stronger privacy preservation.

Evaluating the utility of synthetic COVID-19 case data.

BackgroundConcerns about patient privacy have limited access to COVID-19 datasets. Data synthesis is one approach for making such data broadly available to the research community in a privacy protective manner.ObjectivesEvaluate the utility of synthetic data by comparing analysis results between real and synthetic data.MethodsA gradient boosted classification tree was built to predict death using Ontario’s 90 514 COVID-19 case records linked with community comorbidity, demographic, and socioeconomic characteristics. Model accuracy and relationships were evaluated, as well as privacy risks. The same model was developed on a synthesized dataset and compared to one from the original data.ResultsThe AUROC and AUPRC for the real data model were 0.945 [95% confidence interval (CI), 0.941–0.948] and 0.34 (95% CI, 0.313–0.368), respectively. The synthetic data model had AUROC and AUPRC of 0.94 (95% CI, 0.936–0.944) and 0.313 (95% CI, 0.286–0.342) with confidence interval overlap of 45.05% and 52.02% when compared with the real data. The most important predictors of death for the real and synthetic models were in descending order: age, days since January 1, 2020, type of exposure, and gender. The functional relationships were similar between the two data sets. Attribute disclosure risks were 0.0585, and membership disclosure risk was low.ConclusionsThis synthetic dataset could be used as a proxy for the real dataset.

An efficient clustering-based anonymization scheme for privacy-preserving data collection in IoT based healthcare services

The healthcare services industry has seen a huge transformation since the prominent rise of the Internet of Things (IoT). IoT in healthcare services includes a large number of unified and interconnected sensors, and medical devices that generate and exchange sensitive information. Thus, an enormous amount of data is transmitted through the network which raises an alarming concern for the privacy of patient information. Therefore, privacy preserving data collection (PPDC) is on-demand to ensure the privacy of patient data. Several pieces of research on PPDC have been proposed recently. However, the research literatures have fallen short in privacy requirements and are prone to various privacy attacks. In this paper, we propose a novel privacy-preserving data collection scheme for IoT based healthcare services systems. A clustering-based anonymity model is utilized to develop an efficient privacy-preserving scheme to meet privacy requirements and to prevent healthcare IoT from various privacy attacks. We formulated the threat model as client-server-to-user to ensure privacy on both ends. On the client-side, a modified clustering-based k-anonymity model with α-deassociation is used to anonymize the data generated from the IoT nodes. The base-level privacy is then ensured through a bottom-up clustering method which generates clusters of records as per the privacy requirements. On the server-side, the cluster-combination method-UPGMA is utilized to reduce communication costs and to achieve a better level of privacy. The proposed scheme is efficient in tackling privacy attacks such as attribute disclosure, identity disclosure, membership disclosure, sensitivity attacks, similarity attacks, and skewness attacks. The effectiveness and efficiency of the proposed scheme are proven through theoretical and experimental analyses.

Secure Data Protection Using Slicing as a Confusion Technique

Data Mining deals with automatic extraction of previously unknown patterns from large amounts of data sets. These data sets typically contain sensitive individual information or critical business information, which consequently get exposed to the other parties during Data Mining activities. Secure data protection has been one of the greater concerns in data mining. Several anonymization techniques, such as generalization and bucketization, have been designed for privacy protective microdata publishing. The generalization loses considerable amount of information, especially for high dimensional data. Bucketization, on the other hand, does not prevent membership disclosure and does not apply for data that do not have a clear separation between quasi-identifying attributes and sensitive attributes. Solution to this problem is provided by we introduce a novel data anonymization technique called slicing to improve the current state of the art.

Managing dimensionality in data privacy anonymization

The curse of dimensionality has remained a challenge for a wide variety of algorithms in data mining, clustering, classification, and privacy. Recently, it was shown that an increasing dimensionality makes the data resistant to effective privacy. The theoretical results seem to suggest that the dimensionality curse is a fundamental barrier to privacy preservation. However, in practice, we show that some of the common properties of real data can be leveraged in order to greatly ameliorate the negative effects of the curse of dimensionality. In real data sets, many dimensions contain high levels of inter-attribute correlations. Such correlations enable the use of a process known as vertical fragmentation in order to decompose the data into vertical subsets of smaller dimensionality. An information-theoretic criterion of mutual information is used in the vertical decomposition process. This allows the use of an anonymization process, which is based on combining results from multiple independent fragments. We present a general approach, which can be applied to the k-anonymity, $$\ell $$l-diversity, and t-closeness models. In the presence of inter-attribute correlations, such an approach continues to be much more robust in higher dimensionality, without losing accuracy. We present experimental results illustrating the effectiveness of the approach. This approach is resilient enough to prevent identity, attribute, and membership disclosure attack.

A REVIEW ON ANONYMIZATION TECHNIQUES FOR PRIVACY PRESERVING DATA PUBLISHING

With the increased and vast use of online data, privacy in data publishing has now become very important issue. Now days, many organizations are collecting and storing huge volumes of data in large databases. Data publisher collected data from data holders then data publisher released this data to data recipient for research analysis and mining purpose. The released data reveals some information, which is considered to be private and personal. Privacy of information in such scenario becomes subject of research. In recent years, data anonymization techniques become subject of research. In this paper, we provide a review of the statistical Anonymization techniques that can be used for preserving privacy of publish data. Microdata publishing consist of variety of different anonymization methods like Generalization, Bucketization and Suppression. But for high dimensional data generalization loses significant amount of data and affected from curse of dimensionality. Whereas, bucketization fails to preserve membership disclosure and it needs a clear difference between quasi attributes and Sensitive attributes. Suppression reduces quality of data drastically. To deal with these problems, slicing method is introduced. Slicing is new anonymization technique for preserving publish data. In this data is partitioning horizontally as well as vertically. Dimensionality of the data is decreased by slicing. Utility is preserved & correlations between attributes which are highlycorrelated together are maintained. Compare to generalization, slicing provides better utility of data. Compare with bucketization, slicing is more effective. High-dimensional data can be handled by slicing. Slicing also Provides attribute disclosure and membership disclosure.

Managing Privacy of Sensitive Attributes Using MFSARNN Clustering with Optimization Technique

Several agencies, institutions and organizations publish the sensitive data for the public to achieve better research result.Privacy preserving and security in data publishing are the major challenge.Thus the focus of this paper is to ensure privacy and security of sensitive data by slicing algorithm. The slicing algorithm partition the data into vertical and horizontal columns. The attributes and tuples in the slicing algorithm is clustered based on their similarity. In vertical partitioning, the attributes are grouped by Modified Fully Self Adaptive Resonance Neural Networks (MFSARNN). The cluster formation has been improved by Genetic Algorithm based feature selection. In the horizontal partition, the tuple sare grouped by Metaheuristic Fireflies Algorithm with Minkowsi Distance Measure (MFAMD). In this way the proposed system overcomes the privacy threats such as Identity disclosure, Attribute disclosure and Membership disclosure. The experimental result with respect to mean square error, clustering rate is used to analyse the privacy of the system.

English

Privacy preserving data mining techniques helps in providing security to sensitive information from unauthorized access. Large amount of data is collected in many organizations through data mining. So privacy of data becomes the most important issue in the recent years. Several numbers of techniques such as generalization, bucketization, anonymization have been proposed for privacy preserving data publishing. Generalization loses significant amount of information especially for high-dimensional data according to recent works. Whereas bucketization does not prevent the membership disclosure and cannot applicable to data that does not have clear separation between quasi-identifiers and sensitive attributes. In this paper, we present a slicing technique to prevent generalized loses and membership disclosure. It can also handle high –dimensional data and develops efficient algorithm for computing the sliced data that obeys the ? -diversity check requirement. Slicing preserves better utility than generalization and is more effective than bucketization in workloads involving the sensitive attribute in our experiment

Privacy Preserving Data Publishing through Slicing

Microdata publishing should be privacy preserved as it may contain some sensitive information about an individual. Various anonymization techniques, generalization and bucketization, have been designed for privacy preserving microdata publishing. Generalization does not work better for high dimensional data. Bucketization failed to prevent membership disclosure and does not show a clear separation between quasi-identifiers and sensitive attributes. There are number of attributes in each record which can be categorized as 1) Identifiers such as Name or Social Security Number are the attributes that can be uniquely identify the individuals. 2)Some attributes may be Sensitive Attributes(SAs) such as disease and salary and 3) Some may be Quasi Identifiers (QI) such as zipcode, age, and sex whose values, when taken together, can potentially identify an individual. Data anonymization enables the transfer of information across a boundary, such as between two departments within an agency or between two agencies, while reducing the risk of unintended disclosure, and in certain environments in a manner that enables evaluation and analytics post-anonymization. Here, we present a novel technique called slicing which partitions the data both horizontally and vertically. It preserves better data utility than generalization and is more effective than bucketization in terms of sensitive attribute.

English

Today, most enterprises are actively collecting and storing data in large databases. Many of them have recognized the potential value of these data as an information source for making business decisions. Privacy-preserving data publishing (PPDP) provides methods and tools for publishing useful information while preserving data privacy. In this paper, a brief yet systematic review of several Anonymization techniques such as generalization and Bucketization, have been designed for privacy preserving micro data publishing. Recent work has shown that generalization loses considerable amount of information, especially for high-dimensional data. On the other hand, Bucketization does not prevent membership disclosure. Whereas slicing preserves better data utility than generalization and also prevents membership disclosure. This paper focuses on effective method that can be used for providing better data utility and can handle high dimensional data.

Secure Access to High-dimensional Data through Slicing using Grouping Algorithm

The individual data may be altered, for a variety of purposes. To overcome these concerns, a number of techniques have recently been proposed. Preserving utility of data and actual data from generalization and bucketization in workload involving the sensitive attributes the new technique introduced 'Slicing'. Slicing can handle high dimensional data by partitioning the data sets horizontally and vertically. In slicing data can be organized arbitrarily, checking privacy threats is a concern. Due to the large size of the data sources having several hundred millions to several billions records, and continuously growing, efficient techniques and algorithms are needed. Slicing preserves better data utility than generalization and also prevents membership disclosure. One approach to speed up the processing is to use a process, where potential candidate records are grouped together one and each group is further processed and analyzed on overlapping attributes. The record grouping problem is a formal formulation is to be done in step one. The significance of using slicing is that it can handle high dimension data. Slicing technique used random rows and columns which not give better accuracy hence the new technique of grouping, which improve the working efficiency and accuracy. This paper focus on effective method that can be used for providing better data utility . It can handle high-dimensional data for better security.

A Novel Framework for Privacy Conserving Data Publishing and Handling High Dimensional Data

Now a day’s Data publishing process is the compulsory for visualizing the data sets to other parties. But in publishing process there is thread for data set owners by disclosing the sensitive information. For avoiding those problems we using anonymization techniques for secure data publishing. Most Anonymization techniques such as Generalization and Bucketization are using now. But these anonymization techniques have some limitation. Generalization for k-anonymity losses considerable amount of information for high- dimensional data and Bucketization does not prevent membership disclosure, because bucketization publishes the QI values in their original forms, It requires a clear separation between QI’s and SA’s, but in many data sets, it is unclear which attributes are QI’s and which are SA’s and it breaks the attributes correlations between the QI’s and the SA’s. This paper introduces new anonymization technique slicing to overcome all the drawbacks of bucketization and generalization Keywords: - Sensitive information, High dimensional data, data anonymization, data publishing, data security, generalization and bucketization.

DATA SLICING TECHNIQUE TO PRIVACY PRESERVING AND DATA PUBLISHING

Many techniques have been designed for privacy pres erving and micro data publishing, such as generaliz ation and bucketization. Several works showed that generalization loses some amount of information especially for high dimensio nal data. So it’s not efficient for high dimensional data. In case of Bucketization , it does not prevents membership disclosure and al so does not applicable for data that do not have a clear separation between Quasi-i dentifying attributes and sensitive attributes. In this paper, we presenting an innovative technique called data slicing which part itions the data. An efficient algorithm is develope d for computing sliced data that obeys l-diversity requirement. we also show how dat a slicing is better than generalization and bucketi zation. Data slicing preserves better utility than generalization and also does no t requires clear separation between Quasi-identifyi ng and sensitive attributes. Data slicing is also used to prevent attribute disclosur e and develop an efficient algorithm for computing the sliced data that obeys ldiversity requirement. Experimental results confirm that data slicing preserves data utility than gene ralization and more effective than bucketization involving sensitive attributes. Exper imental results demonstrate the effectiveness of th is method.