Differentially private distributed logistic regression using private and public data

Background

Privacy protecting is an important issue in medical informatics and differential privacy is a state-of-the-art framework for data privacy research. Differential privacy offers provable privacy against attackers who have auxiliary information, and can be applied to data mining models (for example, logistic regression). However, differentially private methods sometimes introduce too much noise and make outputs less useful. Given available public data in medical research (e.g. from patients who sign open-consent agreements), we can design algorithms that use both public and private data sets to decrease the amount of noise that is introduced.

Methodology

In this paper, we modify the update step in Newton-Raphson method to propose a differentially private distributed logistic regression model based on both public and private data.

Experiments and results

We try our algorithm on three different data sets, and show its advantage over: (1) a logistic regression model based solely on public data, and (2) a differentially private distributed logistic regression model based on private data under various scenarios.

Conclusion

Logistic regression models built with our new algorithm based on both private and public datasets demonstrate better utility than models that trained on private or public datasets alone without sacrificing the rigorous privacy guarantee.

     

Will You Accept the Government's Friend Request? Social Networks and Privacy Concerns

Participating in social network websites entails voluntarily sharing private information, and the explosive growth of social network websites over the last decade suggests shifting views on privacy. Concurrently, new anti-terrorism laws, such as the USA Patriot Act, ask citizens to surrender substantial claim to privacy in the name of greater security. I address two important questions regarding individuals'' views on privacy raised by these trends. First, how does prompting individuals to consider security concerns affect their views on government actions that jeopardize privacy? Second, does the use of social network websites alter the effect of prompted security concerns? I posit that prompting individuals to consider security concerns does lead to an increased willingness to accept government actions that jeopardize privacy, but that frequent users of websites like Facebook are less likely to be swayed by prompted security concerns. An embedded survey experiment provides support for both parts of my claim.     

Disclosure of personal medical information: differences among parents and affected adults for genetic and nongenetic conditions

Protecting the confidentiality of medical information has been an issue of great interest in the fields of bioethics, public policy, and law. Few empirical studies have addressed patient experiences and attitudes toward disclosure of private medical information in multiple contexts such as health insurance, employment, and the family. Furthermore, it is unclear whether differences exist in experiences and attitudes about privacy between those living with a serious medical condition versus those who have a child with a medical condition. The study sought to determine whether attitudes and experiences related to medical privacy and confidentiality differ between affected adults and parents of affected children. Interviews were conducted with 296 adults and parents of children with sickle cell disease (SCD), cystic fibrosis (CF), or diabetes mellitus (DM). This cross-sectional study collected data regarding their experiences, attitudes, and beliefs concerning medical privacy and confidentiality. Multinomial logistic regression analysis was conducted on quantitative data. Qualitative analysis was conducted on data from open-ended response items. Parents disclose their child's diagnosis to others more often than affected adults disclose their own disease status. Parents are less likely than affected adults to regret their disclosure, to hope others do not find out, to have been pressured to share information, and to be asked about their disease by employers. Affected adults express greater concern about disclosure, a greater prevalence and greater fear of discrimination, and experience greater pressure from family members to disclose. Clinicians and researchers working with these populations should consider these differences in privacy and disclosure. Further study is necessary to examine the implications of these differences in attitudes and experiences concerning insurance, employment, and social interactions among persons with these conditions.     

Design and Implementation of Website Information Disclosure Assessment System

Internet application technologies, such as cloud computing and cloud storage, have increasingly changed people’s lives. Websites contain vast amounts of personal privacy information. In order to protect this information, network security technologies, such as database protection and data encryption, attract many researchers. The most serious problems concerning web vulnerability are e-mail address and network database leakages. These leakages have many causes. For example, malicious users can steal database contents, taking advantage of mistakes made by programmers and administrators. In order to mitigate this type of abuse, a website information disclosure assessment system is proposed in this study. This system utilizes a series of technologies, such as web crawler algorithms, SQL injection attack detection, and web vulnerability mining, to assess a website’s information disclosure. Thirty websites, randomly sampled from the top 50 world colleges, were used to collect leakage information. This testing showed the importance of increasing the security and privacy of website information for academic websites.     

Information Flow Analysis of Interactome Networks

Recent studies of cellular networks have revealed modular organizations of genes and proteins. For example, in interactome networks, a module refers to a group of interacting proteins that form molecular complexes and/or biochemical pathways and together mediate a biological process. However, it is still poorly understood how biological information is transmitted between different modules. We have developed information flow analysis, a new computational approach that identifies proteins central to the transmission of biological information throughout the network. In the information flow analysis, we represent an interactome network as an electrical circuit, where interactions are modeled as resistors and proteins as interconnecting junctions. Construing the propagation of biological signals as flow of electrical current, our method calculates an information flow score for every protein. Unlike previous metrics of network centrality such as degree or betweenness that only consider topological features, our approach incorporates confidence scores of protein–protein interactions and automatically considers all possible paths in a network when evaluating the importance of each protein. We apply our method to the interactome networks of Saccharomyces cerevisiae and Caenorhabditis elegans. We find that the likelihood of observing lethality and pleiotropy when a protein is eliminated is positively correlated with the protein's information flow score. Even among proteins of low degree or low betweenness, high information scores serve as a strong predictor of loss-of-function lethality or pleiotropy. The correlation between information flow scores and phenotypes supports our hypothesis that the proteins of high information flow reside in central positions in interactome networks. We also show that the ranks of information flow scores are more consistent than that of betweenness when a large amount of noisy data is added to an interactome. Finally, we combine gene expression data with interaction data in C. elegans and construct an interactome network for muscle-specific genes. We find that genes that rank high in terms of information flow in the muscle interactome network but not in the entire network tend to play important roles in muscle function. This framework for studying tissue-specific networks by the information flow model can be applied to other tissues and other organisms as well.     

Computational models of signalling networks for non-linear control

Artificial signalling networks (ASNs) are a computational approach inspired by the signalling processes inside cells that decode outside environmental information. Using evolutionary algorithms to induce complex behaviours, we show how chaotic dynamics in a conservative dynamical system can be controlled. Such dynamics are of particular interest as they mimic the inherent complexity of non-linear physical systems in the real world. Considering the main biological interpretations of cellular signalling, in which complex behaviours and robust cellular responses emerge from the interaction of multiple pathways, we introduce two ASN representations: a stand-alone ASN and a coupled ASN. In particular we note how sophisticated cellular communication mechanisms can lead to effective controllers, where complicated problems can be divided into smaller and independent tasks.     

From single biobanks to international networks: developing e-governance

The future holds the possibility to link and network biobanks, existing biorepositories and reference databases for research purposes in ways that have not been possible before. There is the potential to develop 'research portals' that will enable researchers to access these research resources that are located around the globe with the click of a mouse. In this paper, I will argue that our current governance system for research is unable to provide all of the oversight and accountability mechanisms that are required for this new way of doing research that is based upon flows of data across international borders. For example, our current governance framework for research is nationally based, with a complex system of laws, policies and practice that can be unique to a jurisdiction. It is also evident that many of the nationally based governance bodies in this field do not have the legal powers or expertise to adjudicate on the complex issues, such as privacy and disclosure risks that are raised by cross-border data sharing. In addition, the conceptual underpinning of this research governance structure is based on the "one researcher, one project, one jurisdiction" model. In the conclusion of this paper, I lay out some preliminary ideas as to how this system has to change to accommodate research that is based on networks. I suggest that a move to digital governance mechanisms might be a start to making research governance systems more appropriate for the 21st century.     

Using multilayer network analysis to explore the temporal dynamics of collective behavior

Social organisms often show collective behaviors such as group foraging or movement.Collective behaviors can emerge from interactions between group members and may depend on the behavior of key individuals.When social interactions change over time,collective behaviors may change because these behaviors emerge from interactions among individuals.Despite the importance of,and growing interest in,the temporal dynamics of social interactions,it is not clear how to quantify changes in interactions over time or measure their stability.Furthermore,the temporal scale at which we should observe changes in social networks to detect biologically meaningful changes is not always apparent.Here we use multilayer network analysis to quantify temporal dynamics of social networks of the social spider Stegodyphus dumicola and determine how these dynamics relate to individual and group behaviors.We found that social interactions changed over time at a constant rate.Variation in both network structure and the identity of a keystone individual was not related to the mean or variance of the collective prey attack speed.Individuals that maintained a large and stable number of connections,despite changes in network structure,were the boldest individuals in the group.Therefore,social interactions and boldness are linked across time,but group collective behavior is not influenced by the stability of the social network.Our work demonstrates that dynamic social networks can be modeled in a multilayer framework.This approach may reveal biologically important temporal changes to social structure in other systems.     

Time-ordered networks reveal limitations to information flow in ant colonies

Background

An important function of many complex networks is to inhibit or promote the transmission of disease, resources, or information between individuals. However, little is known about how the temporal dynamics of individual-level interactions affect these networks and constrain their function. Ant colonies are a model comparative system for understanding general principles linking individual-level interactions to network-level functions because interactions among individuals enable integration of multiple sources of information to collectively make decisions, and allocate tasks and resources.

Methodology/Findings

Here we show how the temporal and spatial dynamics of such individual interactions provide upper bounds to rates of colony-level information flow in the ant Temnothorax rugatulus. We develop a general framework for analyzing dynamic networks and a mathematical model that predicts how information flow scales with individual mobility and group size.

Conclusions/Significance

Using thousands of time-stamped interactions between uniquely marked ants in four colonies of a range of sizes, we demonstrate that observed maximum rates of information flow are always slower than predicted, and are constrained by regulation of individual mobility and contact rate. By accounting for the ordering and timing of interactions, we can resolve important difficulties with network sampling frequency and duration, enabling a broader understanding of interaction network functioning across systems and scales.     

A framework to evaluate whether to pool or separate behaviors in a multilayer network

A multilayer network approach combines different network layers,which are connected by interlayer edges,to create a single mathematical object.These networks can contain a variety of information types and represent different aspects of a system.However,the process for selecting which information to include is not always straightforward.Using data on 2 agonistic behaviors in a captive population of monk parakeets(Myiopsitta monachus),we developed a framework for investigating how pooling or splitting behaviors at the scale of dyadic relationships(between 2 individuals)affects individual-and group-level social properties.We designed 2 reference models to test whether randomizing the number of interactions across behavior types results in similar structural patterns as the observed data.Although the behaviors were correlated,the first reference model suggests that the 2 behaviors convey different information about some social properties and should therefore not be pooled.However,once we controlled for data sparsity,we found that the observed measures corresponded with those from the second reference model.Hence,our initial result may have been due to the unequal frequencies of each behavior.Overall,our findings support pooling the 2 behaviors.Awareness of how selected measurements can be affected by data properties is warranted,but nonetheless our framework disentangles these efforts and as a result can be used for myriad types of behaviors and questions.This framework will help researchers make informed and data-driven decisions about which behaviors to pool or separate,prior to using the data in subsequent multilayer network analyses.     

Privacy and disclosure in medical genetics examined in an ethics of care

The progress of genetic knowledge magnifies existing ethical problems in medical genetics. Among the most troubling types of problems -- for medicine, patients, and the larger society -- are those of privacy and disclosure. Examples of the range of problems involving privacy and disclosure are: 1) disclosure of false paternity to an unsuspecting husband; 2) disclosure of a patient's genetic make-up to his or her unknowing spouse; 3) disclosure of information, against a patient's wishes, to relatives at genetic risk; 4) disclosure of ambiguous test results; 5) disclosure of adventitious nonmedical information, e.g., fetal sex; and 6) disclosure to institutional third parties, such as employers and insurers....     

A Spectral Graph Regression Model for Learning Brain Connectivity of Alzheimer’s Disease

Understanding network features of brain pathology is essential to reveal underpinnings of neurodegenerative diseases. In this paper, we introduce a novel graph regression model (GRM) for learning structural brain connectivity of Alzheimer''s disease (AD) measured by amyloid-β deposits. The proposed GRM regards ¹¹C-labeled Pittsburgh Compound-B (PiB) positron emission tomography (PET) imaging data as smooth signals defined on an unknown graph. This graph is then estimated through an optimization framework, which fits the graph to the data with an adjustable level of uniformity of the connection weights. Under the assumed data model, results based on simulated data illustrate that our approach can accurately reconstruct the underlying network, often with better reconstruction than those obtained by both sample correlation and ℓ1-regularized partial correlation estimation. Evaluations performed upon PiB-PET imaging data of 30 AD and 40 elderly normal control (NC) subjects demonstrate that the connectivity patterns revealed by the GRM are easy to interpret and consistent with known pathology. Moreover, the hubs of the reconstructed networks match the cortical hubs given by functional MRI. The discriminative network features including both global connectivity measurements and degree statistics of specific nodes discovered from the AD and NC amyloid-beta networks provide new potential biomarkers for preclinical and clinical AD.     

NET: a new framework for the vectorization and examination of network data

Background

The analysis of complex networks both in general and in particular as pertaining to real biological systems has been the focus of intense scientific attention in the past and present. In this paper we introduce two tools that provide fast and efficient means for the processing and quantification of biological networks like Drosophila tracheoles or leaf venation patterns: the Network Extraction Tool (NET) to extract data and the Graph-edit-GUI (GeGUI) to visualize and modify networks.

Results

NET is especially designed for high-throughput semi-automated analysis of biological datasets containing digital images of networks. The framework starts with the segmentation of the image and then proceeds to vectorization using methodologies from optical character recognition. After a series of steps to clean and improve the quality of the extracted data the framework produces a graph in which the network is represented only by its nodes and neighborhood-relations. The final output contains information about the adjacency matrix of the graph, the width of the edges and the positions of the nodes in space. NET also provides tools for statistical analysis of the network properties, such as the number of nodes or total network length. Other, more complex metrics can be calculated by importing the vectorized network to specialized network analysis packages. GeGUI is designed to facilitate manual correction of non-planar networks as these may contain artifacts or spurious junctions due to branches crossing each other. It is tailored for but not limited to the processing of networks from microscopy images of Drosophila tracheoles.

Conclusion

The networks extracted by NET closely approximate the network depicted in the original image. NET is fast, yields reproducible results and is able to capture the full geometry of the network, including curved branches. Additionally GeGUI allows easy handling and visualization of the networks.

     

Tandem mass spectrometry: a novel approach for metabolic flux analysis

The goal of metabolic flux analysis (MFA) is the accurate estimation of intracellular fluxes in metabolic networks. Here, we introduce a new method for MFA based on tandem mass spectrometry (MS) and stable-isotope tracer experiments. We demonstrate that tandem MS provides more labeling information than can be obtained from traditional full scan MS analysis and allows estimation of fluxes with better precision. We present a modeling framework that takes full advantage of the additional labeling information obtained from tandem MS for MFA. We show that tandem MS data can be computed for any network model, any compound and any tandem MS fragmentation using linear mapping of isotopomers. The inherent advantages of tandem MS were illustrated in two network models using simulated and literature data. Application of tandem MS increased the observability of the models and improved the precision of estimated fluxes by 2- to 5-fold compared to traditional MS analysis.     

Gene regulatory network modeling via global optimization of high-order dynamic Bayesian network

ABSTRACT: BACKGROUND: Dynamic Bayesian network (DBN) is among the mainstream approaches for modeling various biological networks, including the gene regulatory network (GRN). Most current methods for learning DBN employ either local search such as hill-climbing, or a meta stochastic global optimization framework such as genetic algorithm or simulated annealing, which are only able to locate sub-optimal solutions. Further, current DBN applications have essentially been limited to small sized networks. RESULTS: To overcome the above difficulties, we introduce here a deterministic global optimization based DBN approach for reverse engineering genetic networks from time course gene expression data. For such DBN models that consist only of inter time slice arcs, we show that there exists a polynomial time algorithm for learning the globally optimal network structure. The proposed approach, named GlobalMIT+, employs the recently proposed information theoretic scoring metric named mutual information test (MIT). GlobalMIT+ is able to learn high-order time delayed genetic interactions, which are common to most biological systems. Evaluation of the approach using both synthetic and real data sets, including a 733 cyanobacterial gene expression data set, shows significantly improved performance over other techniques. CONCLUSIONS: Our studies demonstrate that deterministic global optimization approaches can infer large scale genetic networks.     

Structural Properties and Complexity of a New Network Class: Collatz Step Graphs

In this paper, we introduce a biologically inspired model to generate complex networks. In contrast to many other construction procedures for growing networks introduced so far, our method generates networks from one-dimensional symbol sequences that are related to the so called Collatz problem from number theory. The major purpose of the present paper is, first, to derive a symbol sequence from the Collatz problem, we call the step sequence, and investigate its structural properties. Second, we introduce a construction procedure for growing networks that is based on these step sequences. Third, we investigate the structural properties of this new network class including their finite scaling and asymptotic behavior of their complexity, average shortest path lengths and clustering coefficients. Interestingly, in contrast to many other network models including the small-world network from Watts & Strogatz, we find that CS graphs become ‘smaller’ with an increasing size.     

Combining Epidemiological and Genetic Networks Signifies the Importance of Early Treatment in HIV-1 Transmission

Inferring disease transmission networks is important in epidemiology in order to understand and prevent the spread of infectious diseases. Reconstruction of the infection transmission networks requires insight into viral genome data as well as social interactions. For the HIV-1 epidemic, current research either uses genetic information of patients'' virus to infer the past infection events or uses statistics of sexual interactions to model the network structure of viral spreading. Methods for a reliable reconstruction of HIV-1 transmission dynamics, taking into account both molecular and societal data are still lacking. The aim of this study is to combine information from both genetic and epidemiological scales to characterize and analyse a transmission network of the HIV-1 epidemic in central Italy.We introduce a novel filter-reduction method to build a network of HIV infected patients based on their social and treatment information. The network is then combined with a genetic network, to infer a hypothetical infection transmission network. We apply this method to a cohort study of HIV-1 infected patients in central Italy and find that patients who are highly connected in the network have longer untreated infection periods. We also find that the network structures for homosexual males and heterosexual populations are heterogeneous, consisting of a majority of ‘peripheral nodes’ that have only a few sexual interactions and a minority of ‘hub nodes’ that have many sexual interactions. Inferring HIV-1 transmission networks using this novel combined approach reveals remarkable correlations between high out-degree individuals and longer untreated infection periods. These findings signify the importance of early treatment and support the potential benefit of wide population screening, management of early diagnoses and anticipated antiretroviral treatment to prevent viral transmission and spread. The approach presented here for reconstructing HIV-1 transmission networks can have important repercussions in the design of intervention strategies for disease control.     

Emergence of Blind Areas in Information Spreading

Recently, contagion-based (disease, information, etc.) spreading on social networks has been extensively studied. In this paper, other than traditional full interaction, we propose a partial interaction based spreading model, considering that the informed individuals would transmit information to only a certain fraction of their neighbors due to the transmission ability in real-world social networks. Simulation results on three representative networks (BA, ER, WS) indicate that the spreading efficiency is highly correlated with the network heterogeneity. In addition, a special phenomenon, namely Information Blind Areas where the network is separated by several information-unreachable clusters, will emerge from the spreading process. Furthermore, we also find that the size distribution of such information blind areas obeys power-law-like distribution, which has very similar exponent with that of site percolation. Detailed analyses show that the critical value is decreasing along with the network heterogeneity for the spreading process, which is complete the contrary to that of random selection. Moreover, the critical value in the latter process is also larger than that of the former for the same network. Those findings might shed some lights in in-depth understanding the effect of network properties on information spreading.     

Sex-biased parasitism,host mass and mutualistic bat flies: an antagonistic individual-based network of bat-bat fly interactions

Individual-based networks provide the building blocks for community-level networks. However, network studies of bats and their parasites have focused only on the species level. Intrapopulation variation may allow certain host individuals to play important roles in the dynamics of the parasites. Therefore, we evaluated how the variation in host sex, body size, ectoparasite abundance and co-occurrence configure individual-based networks of the lesser bulldog bat Noctilio albiventris and bat flies. We expected bat individuals with greater body mass and forearms acting as the core in the network. We also expected males to play a more important role in the network. We sampled a network of N. albiventris bat individuals and their bat flies to describe the structure of an antagonistic individual-based network. We aimed to identify the most relevant bat individuals in the network, focusing on the implications inherent to each of the following approaches: (i) core-periphery organization; (ii) modularity; (iii) species level metrics; and (iv) the main ecological driver of bat individual roles in the network, using niche-based predictors (body mass, forearm and sex). We showed that a network of N. albiventris individuals and their bat flies had low modularity containing a persistent nucleus of individuals and bat flies with well-established interactions. Male individuals with greater body mass played an important role in the network, while for females neither mass nor forearm length were important predictors of their role in the network. Finally, individuals with a high abundance of Paradyschiria parvula played a core role. These results provide an alternative perspective to understand the patterns and mechanisms of interspecific interactions between parasites on the host, as well as sex-biased parasitism.     

A Common Data Set and Framework for Representing Spatial Location Information in the Internet

Many organizations are currently working on how to express and provide location information to services and applications in the Internet. Each of them basically specifies their own way. This raises a problem – the various location information formats, services and applications will not be interoperable in the Internet. Interoperability can be achieved if there is a common way of expressing location information. This paper therefore proposes a common data set and an extensible framework of expressing location information in the Internet. The design aims at bridging various existing/proposed location data representation formats, as well as meeting the requirements of existing/proposed location-aware services.