Data Holdings Research Articles

Machine Unlearning by Reversing the Continual Learning

Recent legislations, such as the European General Data Protection Regulation (GDPR), require user data holders to guarantee the individual’s right to be forgotten. This means that user data holders must completely delete user data upon request. However, in the field of machine learning, it is not possible to simply remove these data from the back-end database wherein the training dataset is stored, because the machine learning model still retains this data information. Retraining the model using a dataset with these data removed can overcome this problem; however, this can lead to expensive computational overheads. In order to remedy this shortcoming, we propose two effective methods to help model owners or data holders remove private data from a trained model. The first method uses an elastic weight consolidation (EWC) constraint term and a modified loss function to neutralize the data to be removed. The second method approximates the posterior distribution of the model as a Gaussian distribution, and the model after unlearning is computed by decreasingly matching the moment (DMM) of the posterior distribution of the neural network trained on all data and the data to be removed. Finally, we conducted experiments on three standard datasets using backdoor attacks as the evaluation metric. The results show that both methods are effective in removing backdoor triggers in deep learning models. Specifically, EWC can reduce the success rate of backdoor attacks to 0. IMM can ensure that the model prediction accuracy is higher than 80% and keep the success rate of backdoor attacks below 10%.

EHR-Safe: generating high-fidelity and privacy-preserving synthetic electronic health records

Privacy concerns often arise as the key bottleneck for the sharing of data between consumers and data holders, particularly for sensitive data such as Electronic Health Records (EHR). This impedes the application of data analytics and ML-based innovations with tremendous potential. One promising approach for such privacy concerns is to instead use synthetic data. We propose a generative modeling framework, EHR-Safe, for generating highly realistic and privacy-preserving synthetic EHR data. EHR-Safe is based on a two-stage model that consists of sequential encoder-decoder networks and generative adversarial networks. Our innovations focus on the key challenging aspects of real-world EHR data: heterogeneity, sparsity, coexistence of numerical and categorical features with distinct characteristics, and time-varying features with highly-varying sequence lengths. Under numerous evaluations, we demonstrate that the fidelity of EHR-Safe is almost-identical with real data (<3% accuracy difference for the models trained on them) while yielding almost-ideal performance in practical privacy metrics.

Population-level benefits of increasing influenza vaccination uptake among Italian older adults: results from a granular panel model.

The impact of seasonal influenza vaccination (SIV) on mortality is still controversial; some studies have claimed that increasing vaccination coverage rates is beneficial, while others have found no significant association. This study aimed to construct a granular longitudinal dataset of local VCRs and assess their effect on pneumonia- and influenza-related (P&I) mortality among Italian adults aged ≥ 65 years. NUTS-3 (nomenclature of territorial units for statistics) level data on SIV coverage were collected via a survey of local data holders. Fixed- and random-effects panel regression modeling, when adjusted for potential confounders, was performed to assess the association between local SIV coverage rates and P&I mortality in older adults. A total of 1,144 local VCRs from 2003 to 2019 were ascertained. In the fully adjusted fixed-effects model, each 1% increase in vaccination coverage was associated (P < 0.001) with a 0.6% (95% CI: 0.3-0.9%) average over-time decrease in P&I mortality. With an annual average of 9,293 P&I deaths in Italy, this model suggested that 56 deaths could have been avoided each year by increasing SIV coverage by 1%. The random-effects model produced similar results. The base-case results were robust in a sensitivity analysis. Over the last two decades, Italian jurisdictions with higher SIV uptake had, on average, fewer P&I deaths among older adults. Local policy-makers should implement effective strategies to increase SIV coverage in the Italian senior population.

Multi-Key Homomorphic Encryption Scheme with Multi-Output Programmable Bootstrapping

Multi-key Homomorphic Encryption (MKHE) scheme can homomorphically evaluate ciphertexts encrypted by different keys, which can effectively protect the privacy information of data holders in the joint computing of cloud services. Since the first full Homomorphic encryption scheme was proposed, bootstrapping is the only way to realize the arbitrary depth homomorphic computation of MKHE schemes. But bootstrap operation is quite expensive. In order to implement fast bootstrapping in MKHE schemes, previous works proposed multi-key TFHE schemes to implement low-latency bootstrapping and output a univariate function of messages after bootstrapping, called Programmable Bootstrapping (PBS). However, these schemes can only encrypt single-bit messages. PBS only outputs a function. And after a homomorphic operation, a bootstrap is required, which undoubtedly results in an increase in the cost of the whole multi-key homomorphic encryption operation. In this paper, we propose a MKHE scheme for multi-output PBS. For this purpose, we study the encryption method and homomorphic operation steps of MKHE, and add BFV homomorphic encryption multiplication and multi-key ciphertext relinearization. We separate the homomorphic operation from bootstrapping. We homomorphically evaluate test polynomials for multiple functions. In contrast to previous MKHE schemes, we support the output of multiple message-related functions with a single bootstrapping operation on the ciphertext. It is no longer limited to encrypting single-bit plaintext, and an effective ciphertext packaging technology is added. According to the analysis given in this paper, it is known that in the scenario of multi-party joint computation, the proposed scheme can be implemented with less bootstrapping when the same number of functions are homomorphically operated. This will effectively reduce the computational overhead.

Big data analysis and changes in customer preferences in post-pandemic global tourism

The list of leading big data holders, whose market intelligence has the most authority for forecasting and planning in global tourism, as well as the reports based on big data analysis are of high importance to be verified. The study reveals a lack of attention to the big data potential for forecasting in tourism and use of traditional limited methods of consumer preferences research published in the annual reports and then circulate in media. Nevertheless, big data analysis implemented by various companies reflects a dramatic change in consumer priorities of the post-pandemic period and captures the demand for authenticity, variety, increased concern for physical and mental health, intangible luxury, and, in general, understanding that traveling is beginning to be perceived as one of essentials of healthy lifestyle. Given these limitations, the list of consumer preferences based on the reported tourism trends for 2023 is of practical relevance to all stakeholders in the industry.

Survey Consent to Administrative Data Linkage: Five Experiments on Wording and Format

Abstract To maximize the value of the data while minimizing respondent burden, survey data are increasingly linked to administrative records. Record linkage often requires the informed consent of survey respondents and failure to obtain consent reduces sample size and may lead to selection bias. Relatively little is known about how best to word and format consent requests in surveys. We conducted a series of experiments in a probability household panel and an online access panel to understand how various features of the design of the consent request can affect informed consent. We experimentally varied: (i) the readability of the consent request, (ii) placement of the consent request in the survey, (iii) consent as default versus the standard opt-in consent question, (iv) offering additional information, and (v) a priming treatment focusing on trust in the data holder. For each experiment, we examine the effects of the treatments on consent rates, objective understanding of the consent request (measured with knowledge test questions), subjective understanding (how well the respondent felt they understood the request), confidence in their decision, response times, and whether they read any of the additional information materials. We find that the default wording and offering additional information do not increase consent rates. Improving the readability of the consent question increases objective understanding but does not increase the consent rate. However, asking for consent early in the survey and priming respondents to consider their trust in the administrative data holder both increase consent rates without negatively affecting understanding of the request.

An Effective and Differentially Private Protocol for Secure Distributed Cardinality Estimation

Counting the number of distinct elements distributed over multiple data holders is a fundamental problem with many real-world applications ranging from crowd counting to network monitoring. Although a number of space and computationally efficient sketch methods (e.g., the Flajolet-Martin sketch and the HyperLogLog sketch) for cardinality estimation have been proposed to solve the above problem, these sketch methods are insecure when considering privacy concerns related to the use of each data holder's personal dataset. Despite a recently proposed protocol that successfully implements the well-known Flajolet-Martin (FM) sketch on a secret-sharing based multiparty computation (MPC) framework for solving the problem of private distributed cardinality estimation (PDCE), we observe that this MPC-FM protocol is not differentially private. In addition, the MPC-FM protocol is computationally expensive, which limits its applications to data holders with limited computation resources. To address the above issues, in this paper we propose a novel protocol DP-DICE, which is computationally efficient and differentially private for solving the problem of PDCE. Experimental results show that our DP-DICE achieves orders of magnitude speedup and reduces the estimation error by several times in comparison with state-of-the-arts under the same security requirements.

Empowering trusted data sharing for data analytics in a federated environment: A blockchain-based approach

As data analytics is used in business to increase profits, organizations use it to pursue their goals. Even if enterprise data could be already valuable on its own, in many cases, combining it with external data sources would boost the value of the output, making data sharing a need in data analytics. At the same time, organizations are reluctant to share data, as they are scared of disclosing critical information. This calls for solutions that are able to safeguard data holders by regulating how data can be shared to ensure the so-called data sovereignty. This paper focuses on the usage of data lakes as well-established technology across enterprises for data analytics where internal or publicly available data are considered. The goal is to extend data lakes with functionalities that, respecting the data sovereignty, enable a data lake also to be ingested with data shared by other organizations and to share data to external organizations. Notable, the purpose of this work is to face this issue by defining an architecture that, inserted in a federated environment: restricts data access and enables monitoring that the actual usage of data respects the data sovereignty expressed in the policies agreed upon by the involved parties; makes use of Blockchain technology as a means for guaranteeing the traceability of data sharing; and allows for balancing computation movement and data movement. The proposed approach has been applied to a healthcare scenario where several institutions (e.g., hospitals and clinics, research institutes, and medical universities) produce and collect clinical data in local data lakes.

Data protection and personal data protection

In Angola, the right to privacy is one of the fundamental rights enshrined in the Constitution of the Republic. In addition to the Constitution, the legal framework on the protection of personal data is complemented by other laws, with regard to the Personal Data Protection Law, No. 22/11 of 17 June, which establishes legal rules for the processing of such data and establishes Data Protection (DPA), as the public authority responsible for monitoring compliance with those rules and exercising. The unequivocal and express consent of the data subject or his legal representative and prior notification to the APD are the general requirements for the processing of personal data, without prejudice to the principles and other specific requirements applicable to sensitive data. Only health professionals registered in the respective professional associations are qualified to process health data, and must observe professional secrecy. Data holders are assured the exercise of their rights to information, access, rectification, updating and deletion of their data. Security measures must be implemented to ensure confidentiality, integrity and availability of information. In conclusion, in Angola there is no prohibition on the processing of personal data, what the legislator imposes is that it be done with transparency and legality, aiming at protecting its holders from situations of vulnerability, including stigmatization, discrimination, observing dignity of the human person as one of the fundamental principles of ethics, especially in scientific research.

Regulating the digital economy: explaining heterogenous business preferences in data governance

ABSTRACT As part of its European Data Strategy, the EU wants to create a market for non-personal data and unlock the sharing of industrial data between companies. This theory-building case study suggests that businesses’ data governance preferences are determined by their position in the data value chain and thus whether firms monetise data as data holders (producers) in the upstream or as data reusers (access seekers) in the downstream segment. While the latter favour legal rules for governing data sharing, the former prefer only contractual agreements. The data value chain theory draws on concepts of the role of data in value creation in the digital economy and on case evidence from the automotive and financial sectors, which are typical cases of digital disruption. The theory helps to better understand coalition building in business lobbying activities, the potential dynamics between private self-regulation and public policies, and the probable feedback effects of digital policies on corporate power sources.

A Data Management Framework for Nurses Using E-Health as a Service (eHaaS)

The electronic health record (EHR) is a patient care database, which helps doctors or nurses to analyse comprehensive patient healthcare through health-cart (h-cart) assistance. Electronic health (e-Health) services offer efficient sharing of the patient's information based on geo-location in which nurses, doctors, or health care practitioners access the patients, promptly and without time delay in case of emergency. In e-Health services, nurses are considered as the data holder who can store and maintain patient's health records in the cloud h-cart platform to analyses patient's data effectively. Therefore, nurses need to safely share and manage access to data in the healthcare system; this need required prominent solutions. However, data authenticity and response time are considered as challenging characteristics in the e-health care system. Hence, in this paper, an improved e-health service model (IeHSM) has been proposed based on cloud computing technology to improve the data authenticity, reliability, and accessibility time of the healthcare information.

Data Access under the Draft Data Act, Competition Law and the DMA: Opening the Data Treasures for Cowmpetition and Innovation?

Abstract In the ongoing transformation from ‘industrial capitalism’ towards an ‘informational capitalism’, one of the core challenges is for the law to design and enforce an appropriate legal framework for access to and use of data. Focusing on access to data generated by the use of products or online services (product and online services usage data), the aim of this paper is to describe and systematize the core elements of this legal framework in the making and to provide some guidance on how it could be further developed. Access by the users of products or online services to the individual-level data that their usage generates must be distinguished from access to bundled individual-level and aggregated data by third parties. With regard to data co-generators’ access to individual-level product usage data, the Draft Data Act proposes to create new rights to access that will become part of a new private law infrastructure of data rights. These access rights are granted independently of whether the data holder is dominant. With regard to data co-generators’ access to individual-level services usage data, the Digital Markets Act (DMA) establishes rights of access only vis-à-vis gatekeepers within the meaning of the DMA. Third-party undertakings who request access to bundled individual-level or aggregated data can, as of now, only base their claim on Art. 102 of the Treaty on the Functioning of the European Union (TFEU) and – in the case of search data – on Art. 6(11) DMA. A regulatory approach has been proposed as an alternative. This paper strives to systematize this hodgepodge of approaches and discern the broader principles that can guide the legislature in creating and fleshing out the legal framework for data. Ideally, markets for bundled individual level or aggregated data will emerge based on the rights of data co-generators to access individual-level data. For this to happen, data intermediaries will have to play a larger role and succeed in bundling and marketing such data.

Contract-Theory-Based Incentive Mechanism for Federated Learning in Health CrowdSensing

Federated learning (FL) provides an effective solution for multiparty data processing under privacy preserving, and becomes a good choice for crowd intelligence extraction in Health CrowdSensing. The quality of the local model submitted by the data holder determines the quality of the global model in FL, and the quality of the local model depends on the data quantity, data quality, and computing power of the data holder. However, in the process of model training, the data holder will inevitably spend the cost of communication and local model training, and higher quality data acquisition and higher quality local model training require higher cost. Therefore, how to motivate data holders with a large amount of high-quality data and computing power to participate in FL has become an urgent problem to be solved. This article transforms the problem of motivating data holders into an optimization problem of utility from the perspective of maximizing the utility of the data holder, establishes the incentive mechanism based on the Contract Theory, and proves that the optimal strategy set of the data holders reaches Nash Equilibrium. A large number of experiments based on public data sets of UCI and MNIST verify that the incentive mechanism can make the baseline algorithm converge faster, while resisting malicious behaviors, such as free-riding and collusive attacks. Furthermore, the data holder with a large amount of high-quality data and computing power can obtain higher revenue.

Regulating for the Future: The Law’s Enforcement Deficit

Regulating for the Future: The Law’s Enforcement Deficit Orla Lynskey (bio) I. Introduction Digitization has changed the way we operate as individuals and as a society in significant ways, with this change accelerated by the Covid-19 pandemic. Digital technologies are now treated as ‘imperative’ across many spheres of life.1 A by-product of this digitization has been datafication, as formerly analogue phenomena need to be translated into data in order to digitize them. As our daily social, professional, administrative, and civic interactions have taken on a digital dimension, the data generated about these transactions are recorded and analysed for insights. The data generated by your morning coffee purchase might therefore reveal your location, your appetite for early mornings, and your price sensitivity, amongst many other insights. The aggregation of such insights across all your daily activities can create remarkably in-depth profiles of individual behaviour. Repeated across the population, datafication facilitates surveillance at a mass scale. The creation and capture of data therefore has clear implications for society. Most evidently, the very act of capturing our interactions through data has privacy implications and could have a chilling effect on the exercise of other rights, including freedom of expression and association. Datafication therefore impacts upon established fundamental rights in democratic societies, with consequences for individuals and society at large. Data also act as a source of power to their holders. Data analytics, including automated decision-making and machine learning, enable the holders of data to categorize and sort individuals in novel ways and to differentiate between cohorts on this basis. While we browse the Internet, the advertisements displayed are targeted based on our profiles, with individuals being grouped into categories to deliver these ads – categories including ‘fertility’, ‘eating disorders’, ‘left-wing politics’ and ‘right-wing politics’, amongst many others.2 Beyond clear privacy implications, the use of such categorizations to differentiate the content that individuals see and the opportunities they are offered offends against ideals of equality, exacerbates asymmetries of power [End Page 104] and information, and has the capacity to further segment already fragmented societies. These data collections have been a source of economic power for their holders and users, but their use goes beyond commerce. The state also engages in such classification-by-data. In England, for instance, resource-pressured local authorities have relied upon automated recommendations to identify children at risk3 while the eligibility of criminal offenders for rehabilitation programmes is assessed, in part, algorithmically, taking into account information scraped from social media sites.4 Increasingly, we see the state operate in tandem with private-sector actors in ways that obscure the flow of data and render effective accountability more difficult to achieve. The challenge for the law, against this backdrop, is to ensure continued respect for the rule of law and fundamental rights. In this contribution, the efforts of the European Union to date will be evaluated with two significant challenges identified. This paper proceeds by, first, providing an overview of the legislative initiatives introduced to regulate the European ‘data economy’. In subsequent sections, it identifies and discusses two key challenges. Firstly, there is an enforcement deficit – to which Ireland contributes – that limits the effectiveness of these laws and detracts from their legitimacy. This enforcement deficit has ironically led to the push for more legislation to plug perceived gaps, which itself may contribute to enforcement pressures. Secondly, and perhaps more fundamentally, there is an ambiguity in European data and digital policy regarding what objectives we seek to promote. While the overarching aim might be to secure rights-respecting innovation, there is little recognition that these two elements – respect for rights and innovation – might be in tension and no clear indication about how conflicts between efficiencies and rights might be resolved. II. The European approach to digital regulation The European approach to digital regulation has been juxtaposed with that of the US, where a liberal ethos and the constraints of the First Amendment protection of speech have characterized its response to digitization, and China, where emphasis has been placed on promoting home-grown tech giants and on data-gathering for surveillance and commercial purposes.5 In contrast, Europe has taken it upon itself to...

Federated learning with hyper-network—a case study on whole slide image analysis

Federated learning(FL) is a new kind of Artificial Intelligence(AI) aimed at data privacy preservation that builds on decentralizing the training data for the deep learning model. This new technique of data security and privacy sheds light on many critical domains with highly sensitive data, including medical image analysis. Developing a strong, scalable, and precise deep learning model has proven to count on a variety of high-quality data from different centers. However, data holders may not willing to share their data considering the restriction of privacy. In this paper, we approach this challenge with a federated learning paradigm. Specifically, we present a case study on the whole slide image classification problem. At each local client center, a multiple-instance learning classifier is developed to conduct whole slide image classification. We introduce a privacy-preserving federated learning framework based on hyper-network to update the global model. Hyper-network is deployed at the global center that produces the weights of the local network conditioned on its input. In this way, hyper-networks can simultaneously learn a family of the local client networks. Instead of communicating raw data with the local client, only model parameters injected with noise are transferred between the local client and the global model. By using a large scale of whole slide images with only slide-level labels, we mensurated our way on two different whole slide image classification problems. The results demonstrate that our proposed federated learning model based on hyper-network can effectively leverage multi-center data to develop a more accurate model which can be used to classify a whole slide image. Its improvements in terms of over the isolated local centers and the commonly used federated averaging baseline are significant. Code will be available.

Individualized PATE: Differentially Private Machine Learning with Individual Privacy Guarantees

Applying machine learning (ML) to sensitive domains requires privacy protection of the underlying training data through formal privacy frameworks, such as differential privacy (DP). Yet, usually, the privacy of the training data comes at the cost of the resulting ML models' utility. One reason for this is that DP uses one uniform privacy budget epsilon for all training data points, which has to align with the strictest privacy requirement encountered among all data holders. In practice, different data holders have different privacy requirements and data points of data holders with lower requirements can contribute more information to the training process of the ML models. To account for this need, we propose two novel methods based on the Private Aggregation of Teacher Ensembles (PATE) framework to support the training of ML models with individualized privacy guarantees. We formally describe the methods, provide a theoretical analysis of their privacy bounds, and experimentally evaluate their effect on the final model's utility using the MNIST, SVHN, and Adult income datasets. Our empirical results show that the individualized privacy methods yield ML models of higher accuracy than the non-individualized baseline. Thereby, we improve the privacy-utility trade-off in scenarios in which different data holders consent to contribute their sensitive data at different individual privacy levels.

Responding to the coronavirus disease-2019 pandemic with innovative data use: The role of data challenges

Abstract Innovative, responsible data use is a critical need in the global response to the coronavirus disease-2019 (COVID-19) pandemic. Yet potentially impactful data are often unavailable to those who could utilize it, particularly in data-poor settings, posing a serious barrier to effective pandemic mitigation. Data challenges, a public call-to-action for innovative data use projects, can identify and address these specific barriers. To understand gaps and progress relevant to effective data use in this context, this study thematically analyses three sets of qualitative data focused on/based in low/middle-income countries: (a) a survey of innovators responding to a data challenge, (b) a survey of organizers of data challenges, and (c) a focus group discussion with professionals using COVID-19 data for evidence-based decision-making. Data quality and accessibility and human resources/institutional capacity were frequently reported limitations to effective data use among innovators. New fit-for-purpose tools and the expansion of partnerships were the most frequently noted areas of progress. Discussion participants identified building capacity for external/national actors to understand the needs of local communities can address a lack of partnerships while de-siloing information. A synthesis of themes demonstrated that gaps, progress, and needs commonly identified by these groups are relevant beyond COVID-19, highlighting the importance of a healthy data ecosystem to address emerging threats. This is supported by data holders prioritizing the availability and accessibility of their data without causing harm; funders and policymakers committed to integrating innovations with existing physical, data, and policy infrastructure; and innovators designing sustainable, multi-use solutions based on principles of good data governance.

A distributed joint extraction framework for sedimentological entities and relations with federated learning

Sedimentological knowledge graphs can be used to identify natural resources in earth layers, which may help geologists analyze the distribution of oil crude in earth, and therefore locating the oilfield that is unknown. The building of such knowledge graphs mainly counts on the methods of joint extraction for pairwise entities and the corresponding relations on large-scale data. However, the whole sedimentological data is fairly owned by the different parties with the possibly inconsistent format. Centralized processing on sedimentological data as a whole will be either securely or structurally impractical. Therefore, this paper proposes a framework of distributed joint extraction in order to harvest knowledge triplets on distributed sedimentological corpus that are from many disparate sources without data transmission. The experimental studies demonstrate our methods not only approach the previous state-of-the-art but also protect the data privacy and security for data holders.

Privacy-Preserving Asynchronous Vertical Federated Learning Algorithms for Multiparty Collaborative Learning.

The privacy-preserving federated learning for vertically partitioned (VP) data has shown promising results as the solution of the emerging multiparty joint modeling application, in which the data holders (such as government branches, private finance, and e-business companies) collaborate throughout the learning process rather than relying on a trusted third party to hold data. However, most of the existing federated learning algorithms for VP data are limited to synchronous computation. To improve the efficiency when the unbalanced computation/communication resources are common among the parties in the federated learning system, it is essential to develop asynchronous training algorithms for VP data while keeping the data privacy. In this article, we propose an asynchronous federated stochastic gradient descent (AFSGD-VP) algorithm and its two variance reduction variants, including stochastic variance reduced gradient (SVRG) and SAGA on the VP data. Moreover, we provide the convergence analyses of AFSGD-VP and its SVRG and SAGA variants under the condition of strong convexity and without any restrictions of staleness. We also discuss their model privacy, data privacy, computational complexities, and communication costs. To the best of our knowledge, AFSGD-VP and its SVRG and SAGA variants are the first asynchronous federated learning algorithms for VP data with theoretical guarantees. Extensive experimental results on a variety of VP datasets not only verify the theoretical results of AFSGD-VP and its SVRG and SAGA variants but also show that our algorithms have much higher efficiency than the corresponding synchronous algorithms.

Review of data quality assessment frameworks experiences around Europe

Abstract In the first phase of its work, WP6 has explored and synthesized existing knowledge and experiences on data quality assurance frameworks (DQAFs) in the context of cross-border sharing of federated secondary use health data. Our aim was to identify good practices within this area and develop a first set of corresponding recommendations. The recommendations reflect the synthesis of results from three parallel lines of work, each utilizing a different methodology: a) thematic workshops and partner meetings, b) analysis of existing data-sharing initiatives, and c) a scoping literature review. Our analysis of literature and existing health data initiatives suggest that currently deployed DQAFs cannot as such serve as platforms or models for an EHDS-wide data quality solution. There is an agreement in principle, that reliability, relevance, timeliness, coherence, coverage and completeness should be incorporated in a DQAF as measurable dimensions of data quality. We aim towards a working definition of data quality, which will be able to reflect the reality of the data, as well as its fitness for purpose from the perspective of potential users. • We recommend focusing efforts on transparency at the level of data holder institutions across Member States in relation to adoption of regular audits, a well-developed DQAF and clear procedures with regard to processing the data. • In the medium to longer-term EHDS nodes could promote and support the development of a benchmarking process, which will assist data managers and data holder institutions with alignment against a Europe-wide approach to measuring data quality within and across Member States.