Narrow Results

Protecting healthcare data privacy and security is crucial in advanced manufacturing, which involves medical devices. It encompasses patient records and clinical trial data. Federated learning emerges as a solution that enables model training across different institutions without compromising data privacy and security. However, existing frameworks often exhibit a bias towards clients with larger data volumes, neglecting the connection between global and local model performance. This can result in suboptimal aggregation of the global model, thereby affecting the effectiveness and efficiency of the overall process. To address these limitations, we propose a performance evaluation-driven federated learning framework (PedFed). The primary objective of PedFed is to enhance global model aggregation and improve communication efficiency. Our approach involves a client selection strategy based on performance evaluation of local and global models. Specifically, we introduce the concept of local model improvement (LMI) using Intersection over Union (IoU) for client selection in medical image segmentation scenarios. Moreover, we introduce a dynamic aggregation framework incorporating validation IoU as a weighting factor to mitigate model divergence caused by not independent and identically distributed (non-IID) data. We focus on performing image segmentation tasks to simulate the analysis of sensitive data in the healthcare domain. Experimental results conducted on brain tumor and heart segmentation datasets demonstrate the superiority of the PedFed framework over the baseline framework, confirming its benefits in communication efficiency.

In most systems, a smart functionality is enabled through an essential vital service such as detecting anomalies from complex, large-scale and dynamic data. However, ensuring the privacy and security for the cloud data is the most crucial and challenging task in the present world. Moreover, it is important to safeguard the security of sensitive data and its privacy from unauthorized parties who are trying to access the data. Therefore, to accomplish this task, several encryption, decryption and key generation mechanisms were introduced in the existing works for privacy preserving in cloud platform. But, there still remain open issues such as increased communication overhead, reduced security and increased time consumption. Also, these existing works followed the symmetric key cryptographic mechanism for privacy preservation of data; hence, a single secret key is shared by several users for accessing the original data. Due to this fact, a high security risk arises and it allows unauthorized parties to access the data. Thus, this work introduces a cloud-based privacy preserving model for offering a scalable and reliable anomaly detection service for sensor data through holding the benefits of cloud resources. Also, this paper aims to impose a newly developed Elliptic Curve Cryptography-based Collective Decision Optimization (ECDO) approach over the proposed framework for improving the privacy and security of the data. Furthermore, to perform the data clustering computation we used the Gaussian kernel fuzzy $c$-means clustering (GKFCM) algorithm within the cloud platform, especially for data partitioning and to classify the anomalies. Thus, the computational difficulties are limited by adopting this suitable privacy preserving model which collaborates a private server and a set of public servers through a cloud data center. Moreover, on encrypted data the granular anomaly detection operations are performed by the virtual nodes executed over public servers. Experimental validation was performed on four datasets resulting from Intel Labs publicly available sensor data. The experimental outcomes demonstrate the ability of the proposed framework in providing high anomaly detection accuracy without any degradation in data privacy.

Consider a data holder, such as a hospital or a bank, that has a privately held collection of person-specific, field structured data. Suppose the data holder wants to share a version of the data with researchers. How can a data holder release a version of its private data with scientific guarantees that the individuals who are the subjects of the data cannot be re-identified while the data remain practically useful? The solution provided in this paper includes a formal protection model named k-anonymity and a set of accompanying policies for deployment. A release provides k-anonymity protection if the information for each person contained in the release cannot be distinguished from at least k-1 individuals whose information also appears in the release. This paper also examines re-identification attacks that can be realized on releases that adhere to k-anonymity unless accompanying policies are respected. The k-anonymity protection model is important because it forms the basis on which the real-world systems known as Datafly, μ-Argus and k-Similar provide guarantees of privacy protection.

Often a data holder, such as a hospital or bank, needs to share person-specific records in such a way that the identities of the individuals who are the subjects of the data cannot be determined. One way to achieve this is to have the released records adhere to k-anonymity, which means each released record has at least (k-1) other records in the release whose values are indistinct over those fields that appear in external data. So, k-anonymity provides privacy protection by guaranteeing that each released record will relate to at least k individuals even if the records are directly linked to external information. This paper provides a formal presentation of combining generalization and suppression to achieve k-anonymity. Generalization involves replacing (or recoding) a value with a less specific but semantically consistent value. Suppression involves not releasing a value at all. The Preferred Minimal Generalization Algorithm (MinGen), which is a theoretical algorithm presented herein, combines these techniques to provide k-anonymity protection with minimal distortion. The real-world algorithms Datafly and μ-Argus are compared to MinGen. Both Datafly and μ-Argus use heuristics to make approximations, and so, they do not always yield optimal results. It is shown that Datafly can over distort data and μ-Argus can additionally fail to provide adequate protection.

In this paper we review the state of the art on graph privacy with special emphasis on applications to online social networks, and we consider some novel aspects which have not been greatly covered in the specialized literature on graph privacy. The following key considerations are covered: (i) choice of different operators to modify the graph; (ii) information loss based on the cost of graph operations in terms of statistical characteristics (degree, clustering coefficient and path length) in the original graph; (iii) computational cost of the operations; (iv) in the case of the aggregation of two nodes, the choice of similar adjacent nodes rather than isomorphic topologies, in order to maintain the overall structure of the graph; (v) a statistically knowledgeable attacker who is able to search for regions of a simply anonymized graph based on statistical characteristics and map those onto a given node and its immediate neighborhood.

In this paper, we present an overview of p-sensitive k-anonymity models including the basic model, the extended p-sensitive k-anonymity, the constrained p-sensitive k-anonymity, and the (p⁺, α)-sensitive k-anonymity. Existing properties of these models are reviewed and illustrated, and new properties regarding the maximum number of QI-clusters are discussed and proved. This paper includes a review of related anonymity models and a very brief summary of existing algorithms for the family of p-sensitive k-anonymity models.

In data privacy, the evaluation of the disclosure risk has to take into account the fact that several releases of the same or similar information about a population are common. In this paper we discuss this issue within the scope of k-anonymity. We also show how this issue is related to the publication of privacy protected databases that consist of linked tables. We present algorithms for the implementation of k-anonymity for this type of data.

In this paper, we propose an approach for preserving trade secrets in B2B interactions among competitors. Customer information exchanged during Business-to-Business (B2B) collaborations is usually considered as a business asset not to be freely shared with other businesses. This customer information is in essence a business trade secret. The full automation of B2B interactions is now possible because of the wide deployment of such technologies as Web services. However, these advances also create greater opportunities for businesses in acquiring sensitive customer data from other transacting businesses, thus creating an impediment for potential B2B collaboration between competitors. This calls for techniques to protect disclosure of sensitive data. Our approach leverages psycholinguistic knowledge to perturb data to computationally impede the disclosure of privileged customer information. We present an analytical model and a set of experiments to demonstrate the robustness of the proposed techniques.

Explosive growth of information in the Internet has raised threats for individual privacy. k-Anonymity and l-diversity are two known techniques proposed to address the threats. They use concept hierarchy tree (CHT)-based generalization/suppression. For a given attribute several CHTs can be constructed. An appropriate CHT is to be chosen for attribute anonymization to be effective. This paper discusses an on the fly approach for constructing CHT which can be used for generalization/suppression. Furthermore to improve anonymization the CHT can be dynamically adjusted for a given k value. Performance evaluation is done for the proposed approach and a comparative study is performed against known methods, k-member clustering anonymization and mondrian multi-dimensional algorithm using (1) improved on the fly hierarchy (IOTF) (Campan et al., 2011), (2) on the fly hierarchy (OTF) (Campan and Cooper, 2010), (3) hierarchy free (HF) (LeFevre et al., 2006), (4) predefined hierarchy (PH) (Iyengar, 2002) (5) CHU (Chu and Chiang, 1994) and (6) HAN (Han and Fu, 1994) methods. The metrics used for evaluation are (a) information loss, (b) discernibility metric, (c) normalized average equivalence size metric. Experimental results indicate that our approach is more effective and flexible and the utility is 12% better than IOTF, 16% better than OTF and CHU, 17% better than PH and 21% better than HAN methods when applied on mondrain multi-dimensional algorithm. Experiments are conducted on k-member clustering technique and it is observed that our approach improved utility 1% better than IOTF, 2% better than OTF, 3% better than CHU, 5% better than PH and 14% better than HAN methods.

Federated learning (FL), a cutting-edge approach in machine learning, enables model training across decentralized devices while preserving data privacy and reducing communication costs. This paper explores the vast landscape of opportunities and challenges associated with FL. The opportunities lie in its potential to leverage vast, diverse datasets from edge devices, enabling the creation of highly personalized models. This decentralization also promotes privacy compliance, making it ideal for sectors like healthcare and finance, where data security is paramount. Furthermore, FL opens up avenues for collaborative research without centralized data control, fostering innovation in various domains. This paper provides a comprehensive analysis of these opportunities and challenges, offering insights into the future directions of FL research. By understanding these dynamics, researchers, policymakers, and industry practitioners can collaboratively work towards harnessing the full potential of FL while mitigating its inherent challenges.

Anonymized electronic health records (EHR) are often used for biomedical research. One persistent concern with this type of research is the risk for re-identification of patients from their purportedly anonymized data. Here, we use the EHR of 731,850 de-identified patients to demonstrate that the average patient is unique from all others 98.4% of the time simply by examining what laboratory tests have been ordered for them. By the time a patient has visited the hospital on two separate days, they are unique in 72.3% of cases. We further present a computational study to identify how accurately the records from a single day of care can be used to re-identify patients from a set of 99 other patients. We show that, given a single visit’s laboratory orders (even without result values) for a patient, we can re-identify the patient at least 25% of the time. Furthermore, we can place this patient among the top 10 most similar patients 47% of the time. Finally, we present a proof-of-concept technique using a variational autoencoder to encode laboratory results into a lower-dimensional latent space. We demonstrate that releasing latentspace encoded laboratory orders significantly improves privacy compared to releasing raw laboratory orders (<5% re-identification), while preserving information contained within the laboratory orders (AUC of >0.9 for recreating encoded values). Our findings have potential consequences for the public release of anonymized laboratory tests to the biomedical research community. We note that our findings do not imply that laboratory tests alone are personally identifiable. In the attack scenario presented here, reidentification would require a threat actor to possess an external source of laboratory values which are linked to personal identifiers at the start.

There is a disconnect between data practices in biomedicine and public understanding of those data practices, and this disconnect is expanding rapidly every day (with the emergence of synthetic data and digital twins and more widely adopted Artificial Intelligence (AI)/Machine Learning tools). Transparency alone is insufficient to bridge this gap. Concurrently, there is an increasingly complex landscape of laws, regulations, and institutional/ programmatic policies to navigate when engaged in biocomputing and digital health research, which makes it increasingly difficult for those wanting to “get it right” or “do the right thing.” Mandatory data protection obligations vary widely, sometimes focused on the type of data (and nuanced definition and scope parameters), the actor/entity involved, or the residency of the data subjects. Additional challenges come from attempts to celebrate biocomputing discoveries and digital health innovations, which frequently transform fair and accurate communications into exaggerated hype (e.g., to secure financial investment in future projects or lead to more favorable tenure and promotion decisions). Trust in scientists and scientific expertise can be quickly eroded if, for example, synthetic data is perceived by the public as “fake data” or if digital twins are perceived as “imaginary” patients. Researchers appear increasingly aware of the scientific and moral imperative to strengthen their work and facilitate its sustainability through increased diversity and community engagement. Moreover, there is a growing appreciation for the “data work” necessary to have scientific data become meaningful, actionable information, knowledge, and wisdom–not only for scientists but also for the individuals from whom those data were derived or to whom those data relate. Equity in the process of biocomputing and equity in the distribution of benefits and burdens of biocomputing both demand ongoing development, implementation, and refinement of embedded Ethical, Legal and Social Implications (ELSI) research practices. This workshop is intended to nurture interdisciplinary discussion of these issues and to highlight the skills and competencies all too often considered “soft skills” peripheral to other skills prioritized in traditional training and professional development programs. Data scientists attending this workshop will become better equipped to embed ELSI practices into their research.