[Show abstract][Hide abstract] ABSTRACT: Today, many people spend a lot of time online. Their social interactions captured in online social networks are an important part of the overall personal social profile, in addition to interactions taking place offline. This paper investigates whether relations captured by online social networks can be used as a proxy for the relations in offline social networks, such as networks of human face-to-face (F2F) proximity and coauthorship networks. Particularly, the paper focuses on interactions of computer scientists in online settings (homepages, social networks profiles and connections) and offline settings (scientific collaboration, face-to-face communications during the conferences). We focus on quantitative studies and investigate the structural similarities and correlations of the induced networks; in addition, we analyze implications between networks. Finally, we provide a qualitative user analysis to find characteristics of good and bad proxies.
[Show abstract][Hide abstract] ABSTRACT: Data describing human interactions often suffer from incomplete sampling of
the underlying population. As a consequence, the study of contagion processes
using data-driven models can lead to a severe underestimation of the epidemic
risk. Here we present a systematic method to correct this bias and obtain an
accurate estimation of the risk in the context of epidemic models informed by
high-resolution time-resolved contact data. We consider several such data sets
collected in various contexts and perform controlled resampling experiments. We
show that the statistical information contained in the resampled data allows us
to build surrogate versions of the unknown contacts and that simulations of
epidemic processes using these surrogate data sets yield good estimates of the
outcome of simulations performed using the complete data set. We discuss
limitations and potential improvements of our method.
[Show abstract][Hide abstract] ABSTRACT: OBJECTIVE Contact patterns and microbiological data contribute to a detailed understanding of infectious disease transmission. We explored the automated collection of high-resolution contact data by wearable sensors combined with virological data to investigate influenza transmission among patients and healthcare workers in a geriatric unit. DESIGN Proof-of-concept observational study. Detailed information on contact patterns were collected by wearable sensors over 12 days. Systematic nasopharyngeal swabs were taken, analyzed for influenza A and B viruses by real-time polymerase chain reaction, and cultured for phylogenetic analysis. SETTING An acute-care geriatric unit in a tertiary care hospital. PARTICIPANTS Patients, nurses, and medical doctors. RESULTS A total of 18,765 contacts were recorded among 37 patients, 32 nurses, and 15 medical doctors. Most contacts occurred between nurses or between a nurse and a patient. Fifteen individuals had influenza A (H3N2). Among these, 11 study participants were positive at the beginning of the study or at admission, and 3 patients and 1 nurse acquired laboratory-confirmed influenza during the study. Infectious medical doctors and nurses were identified as potential sources of hospital-acquired influenza (HA-Flu) for patients, and infectious patients were identified as likely sources for nurses. Only 1 potential transmission between nurses was observed. CONCLUSIONS Combining high-resolution contact data and virological data allowed us to identify a potential transmission route in each possible case of HA-Flu. This promising method should be applied for longer periods in larger populations, with more complete use of phylogenetic analyses, for a better understanding of influenza transmission dynamics in a hospital setting. Infect Control Hosp Epidemiol 2015;00(0): 1-7.
Infection Control and Hospital Epidemiology 03/2015; 36(3):1-7. DOI:10.1017/ice.2014.53 · 4.18 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: The customary perspective to reason about epidemic mitigation in temporal
networks hinges on the identification of nodes with specific features or
network roles. The ensuing individual-based control strategies, however, are
difficult to carry out in practice and ignore important correlations between
topological and temporal patterns. Here we adopt a mesoscopic perspective and
present a principled framework to identify collective features at multiple
scales and rank their importance for epidemic spread. We use tensor
decomposition techniques to build an additive representation of a temporal
network in terms of mesostructures, such as cohesive clusters and
temporally-localized mixing patterns. This representation allows to determine
the impact of individual mesostructures on epidemic spread and to assess the
effect of targeted interventions that remove chosen structures. We illustrate
this approach using high-resolution social network data on face-to-face
interactions in a school and show that our method affords the design of
effective mesoscale interventions.
[Show abstract][Hide abstract] ABSTRACT: Background
School environments are thought to play an important role in the community spread of infectious diseases such as influenza because of the high mixing rates of school children. The closure of schools has therefore been proposed as an efficient mitigation strategy. Such measures come however with high associated social and economic costs, making alternative, less disruptive interventions highly desirable. The recent availability of high-resolution contact network data from school environments provides an opportunity to design models of micro-interventions and compare the outcomes of alternative mitigation measures.
Methods and results
We model mitigation measures that involve the targeted closure of school classes or grades based on readily available information such as the number of symptomatic infectious children in a class. We focus on the specific case of a primary school for which we have high-resolution data on the close-range interactions of children and teachers. We simulate the spread of an influenza-like illness in this population by using an SEIR model with asymptomatics, and compare the outcomes of different mitigation strategies. We find that targeted class closure affords strong mitigation effects: closing a class for a fixed period of time – equal to the sum of the average infectious and latent durations – whenever two infectious individuals are detected in that class decreases the attack rate by almost 70% and significantly decreases the probability of a severe outbreak. The closure of all classes of the same grade mitigates the spread almost as much as closing the whole school.
Our model of targeted class closure strategies based on readily available information on symptomatic subjects and on limited information on mixing patterns, such as the grade structure of the school, shows that these strategies might be almost as effective as whole-school closure, at a much lower cost. This may inform public health policies for the management and mitigation of influenza-like outbreaks in the community.
Electronic supplementary material
The online version of this article (doi:10.1186/s12879-014-0695-9) contains supplementary material, which is available to authorized users.
[Show abstract][Hide abstract] ABSTRACT: How are social interaction dynamics associated with mental health during early stages of adolescence? The goal of this study is to objectively measure social interactions and evaluate the roles that multiple aspects of the social environment - such as physical activity and food choice - may jointly play in shaping the structure of children's relationships and their mental health. The data in this study are drawn from a longitudinal network-behavior study conducted in 2012 at a private K-8 school in an urban setting in California. We recruited a highly complete network sample of sixth-graders (n = 40, 91% of grade, mean age = 12.3), and examined how two measures of distressed mental health (self-esteem and depressive symptoms) are positionally distributed in an early adolescent interaction network. We ascertained how distressed mental health shapes the structure of relationships over a three-month period, adjusting for relevant dimensions of the social environment. Cross-sectional analyses of interaction networks revealed that self-esteem and depressive symptoms are differentially stratified by gender. Specifically, girls with more depressive symptoms have interactions consistent with social inhibition, while boys' interactions suggest robustness to depressive symptoms. Girls higher in self-esteem tended towards greater sociability. Longitudinal network behavior models indicate that gender similarity and perceived popularity are influential in the formation of social ties. Greater school connectedness predicts the development of self-esteem, though social ties contribute to more self-esteem improvement among students who identify as European-American. Cross-sectional evidence shows associations between distressed mental health and students' network peers. However, there is no evidence that connected students' mental health status becomes more similar in their over time because of their network interactions. These findings suggest that mental health during early adolescence may be less subject to mechanisms of social influence than network research in even slightly older adolescents currently indicates.
Social Science [?] Medicine 04/2014; 125. DOI:10.1016/j.socscimed.2014.04.015 · 2.89 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Streams of user-generated content in social media exhibit patterns of
collective attention across diverse topics, with temporal structures determined
both by exogenous factors and endogenous factors. Teasing apart different
topics and resolving their individual, concurrent, activity timelines is a key
challenge in extracting knowledge from microblog streams. Facing this challenge
requires the use of methods that expose latent signals by using term
correlations across posts and over time. Here we focus on content posted to
Twitter during the London 2012 Olympics, for which a detailed schedule of
events is independently available and can be used for reference. We mine the
temporal structure of topical activity by using two methods based on
non-negative matrix factorization. We show that for events in the Olympics
schedule that can be semantically matched to Twitter topics, the extracted
Twitter activity timeline closely matches the known timeline from the schedule.
Our results show that, given appropriate techniques to detect latent signals,
Twitter can be used as a social sensor to extract topical-temporal information
on real-world events at high temporal resolution.
[Show abstract][Hide abstract] ABSTRACT: The increasing availability of temporal network data is calling for more research on extracting and characterizing mesoscopic structures in temporal networks and on relating such structure to specific functions or properties of the system. An outstanding challenge is the extension of the results achieved for static networks to time-varying networks, where the topological structure of the system and the temporal activity patterns of its components are intertwined. Here we investigate the use of a latent factor decomposition technique, non-negative tensor factorization, to extract the community-activity structure of temporal networks. The method is intrinsically temporal and allows to simultaneously identify communities and to track their activity over time. We represent the time-varying adjacency matrix of a temporal network as a three-way tensor and approximate this tensor as a sum of terms that can be interpreted as communities of nodes with an associated activity time series. We summarize known computational techniques for tensor decomposition and discuss some quality metrics that can be used to tune the complexity of the factorized representation. We subsequently apply tensor factorization to a temporal network for which a ground truth is available for both the community structure and the temporal activity patterns. The data we use describe the social interactions of students in a school, the associations between students and school classes, and the spatio-temporal trajectories of students over time. We show that non-negative tensor factorization is capable of recovering the class structure with high accuracy. In particular, the extracted tensor components can be validated either as known school classes, or in terms of correlated activity patterns, i.e., of spatial and temporal coincidences that are determined by the known school activity schedule.
PLoS ONE 01/2014; 9(1):e86028. DOI:10.1371/journal.pone.0086028 · 3.23 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Thanks to recent technological advances, measuring real-world interactions using mobile devices and wearable sensors has become possible, allowing researchers to gather data on human social interactions in a variety of contexts with a high spatial and temporal resolution. Empirical data describing contact networks reach thus a high level of detail that may yield bring new insights into the dynamics of infection transmission between individuals. At the same time, such data bring forth new challenges related to their statistical description and analysis and to their use in mathematical models. In particular, the integration of highly detailed empirical data in computational frameworks designed to model the spread of infectious diseases raises the issue of assessing which representations of the raw data work best to inform the models. There is an emerging need to strike a balance between simplicity and detail in order to ensure both generalizability and accuracy of predictions. Here, we review recent work on the collection and analysis of highly detailed data on temporal networks of face-to-face human proximity, carried out in the context of the SocioPatterns collaboration. We discuss the various levels of coarse-graining that can be used to represent the data in order to inform models of infectious diseases transmission. We moreover discuss several limitations of the data and future avenues for data collection and modeling efforts in the field of infectious diseases. This article is protected by copyright. All rights reserved.
[Show abstract][Hide abstract] ABSTRACT: Dynamical processes on time-varying complex networks are key to understanding and modeling a broad variety of processes in socio-technical systems. Here we focus on empirical temporal networks of human proximity and we aim at understanding the factors that, in simulation, shape the arrival time distribution of simple spreading processes. Abandoning the notion of wall-clock time in favour of node-specific clocks based on activity exposes robust statistical patterns in the arrival times across different social contexts. Using randomization strategies and generative models constrained by data, we show that these patterns can be understood in terms of heterogeneous inter-event time distributions coupled with heterogeneous numbers of events per edge. We also show, both empirically and by using a synthetic dataset, that significant deviations from the above behavior can be caused by the presence of edge classes with strong activity correlations.
[Show abstract][Hide abstract] ABSTRACT: Contacts between patients, patients and health care workers (HCWs) and among HCWs represent one of the important routes of transmission of hospital-acquired infections (HAI). A detailed description and quantification of contacts in hospitals provides key information for HAIs epidemiology and for the design and validation of control measures.
We used wearable sensors to detect close-range interactions ("contacts") between individuals in the geriatric unit of a university hospital. Contact events were measured with a spatial resolution of about 1.5 meters and a temporal resolution of 20 seconds. The study included 46 HCWs and 29 patients and lasted for 4 days and 4 nights. 14,037 contacts were recorded overall, 94.1% of which during daytime. The number and duration of contacts varied between mornings, afternoons and nights, and contact matrices describing the mixing patterns between HCW and patients were built for each time period. Contact patterns were qualitatively similar from one day to the next. 38% of the contacts occurred between pairs of HCWs and 6 HCWs accounted for 42% of all the contacts including at least one patient, suggesting a population of individuals who could potentially act as super-spreaders.
Wearable sensors represent a novel tool for the measurement of contact patterns in hospitals. The collected data can provide information on important aspects that impact the spreading patterns of infectious diseases, such as the strong heterogeneity of contact numbers and durations across individuals, the variability in the number of contacts during a day, and the fraction of repeated contacts across days. This variability is however associated with a marked statistical stability of contact and mixing patterns across days. Our results highlight the need for such measurement efforts in order to correctly inform mathematical models of HAIs and use them to inform the design and evaluation of prevention strategies.
PLoS ONE 09/2013; 8(9):e73970. DOI:10.1371/journal.pone.0073970 · 3.23 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: The ever increasing adoption of mobile technologies and ubiquitous services allows to sense human behavior at unprecedented level of details and scale. Wearable sensors, in particular, open up a new window on human mobility and proximity in a variety of indoor environments. Here we review stylized facts on the structural and dynamical properties of empirical networks of human face-to-face proximity, measured in three different real-world contexts: an academic conference, a hospital ward, and a museum exhibition. First, we discuss the structure of the aggregated contact networks, that project out the detailed ordering of contact events while preserving temporal heterogeneities in their weights. We show that the structural properties of aggregated networks highlight important differences and unexpected similarities across contexts, and discuss the additional complexity that arises from attributes that are typically associated with nodes in real-world interaction networks, such as role classes in hospitals. We then consider the empirical data at the finest level of detail, i.e., we consider time-dependent networks of face-to-face proximity between individuals. To gain insights on the effects that causal constraints have on spreading processes, we simulate the dynamics of a simple susceptible-infected model over the empirical time-resolved contact data. We show that the spreading pathways for the epidemic process are strongly affected by the temporal structure of the network data, and that the mere knowledge of static aggregated networks leads to erroneous conclusions about the transmission paths on the corresponding dynamical networks.
The European Physical Journal Special Topics 09/2013; 222(6):1295-1309. DOI:10.1140/epjst/e2013-01927-7 · 1.40 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Spreading processes represent a very efficient tool to investigate the structural properties of networks and the relative importance of their constituents, and have been widely used to this aim in static networks. Here we consider simple disease spreading processes on empirical time-varying networks of contacts between individuals, and compare the effect of several immunization strategies on these processes. An immunization strategy is defined as the choice of a set of nodes (individuals) who cannot catch nor transmit the disease. This choice is performed according to a certain ranking of the nodes of the contact network. We consider various ranking strategies, focusing in particular on the role of the training window during which the nodes' properties are measured in the time-varying network: longer training windows correspond to a larger amount of information collected and could be expected to result in better performances of the immunization strategies. We find instead an unexpected saturation in the efficiency of strategies based on nodes' characteristics when the length of the training window is increased, showing that a limited amount of information on the contact patterns is sufficient to design efficient immunization strategies. This finding is balanced by the large variations of the contact patterns, which strongly alter the importance of nodes from one period to the next and therefore significantly limit the efficiency of any strategy based on an importance ranking of nodes. We also observe that the efficiency of strategies that include an element of randomness and are based on temporally local information do not perform as well but are largely independent on the amount of information available.
[Show abstract][Hide abstract] ABSTRACT: We investigate gender homophily in the spatial proximity of children (6 to 12
years old) in a French primary school, using time-resolved data on face-to-face
proximity recorded by means of wearable sensors. For strong ties, i.e., for
pairs of children who interact more than a defined threshold, we find
statistical evidence of gender preference that increases with grade. For weak
ties, conversely, gender homophily is negatively correlated with grade for
girls, and positively correlated with grade for boys. This different evolution
with grade of weak and strong ties exposes a contrasted picture of gender
Social Networks 06/2013; 35(4). DOI:10.1016/j.socnet.2013.08.003 · 2.93 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Representing and efficiently querying time-varying social network data is a central challenge that needs to be addressed in order to support a variety of emerging applications that leverage high-resolution records of human activities and interactions from mobile devices and wearable sensors. In order to support the needs of specific applications, as well as general tasks related to data curation, cleaning, linking, post-processing, and data analysis, data models and data stores are needed that afford efficient and scalable querying of the data. In particular, it is important to design solutions that allow rich queries that simultaneously involve the topology of the social network, temporal information on the presence and interactions of individual nodes, and node metadata. Here we introduce a data model for time-varying social network data that can be represented as a property graph in the Neo4j graph database. We use time-varying social network data collected by using wearable sensors and study the performance of real-world queries, pointing to strengths, weaknesses and challenges of the proposed approach.
First International Workshop on Graph Data Management Experiences and Systems; 06/2013
[Show abstract][Hide abstract] ABSTRACT: The ever increasing adoption of mobile technologies and ubiquitous services
allows to sense human behavior at unprecedented levels of details and scale.
Wearable sensors are opening up a new window on human mobility and proximity at
the finest resolution of face-to-face proximity. As a consequence, empirical
data describing social and behavioral networks are acquiring a longitudinal
dimension that brings forth new challenges for analysis and modeling. Here we
review recent work on the representation and analysis of temporal networks of
face-to-face human proximity, based on large-scale datasets collected in the
context of the SocioPatterns collaboration. We show that the raw behavioral
data can be studied at various levels of coarse-graining, which turn out to be
complementary to one another, with each level exposing different features of
the underlying system. We briefly review a generative model of temporal contact
networks that reproduces some statistical observables. Then, we shift our focus
from surface statistical features to dynamical processes on empirical temporal
networks. We discuss how simple dynamical processes can be used as probes to
expose important features of the interaction patterns, such as burstiness and
causal constraints. We show that simulating dynamical processes on empirical
temporal networks can unveil differences between datasets that would otherwise
look statistically similar. Moreover, we argue that, due to the temporal
heterogeneity of human dynamics, in order to investigate the temporal
properties of spreading processes it may be necessary to abandon the notion of
wall-clock time in favour of an intrinsic notion of time for each individual
node, defined in terms of its activity level. We conclude highlighting several
open research questions raised by the nature of the data at hand.
Understanding Complex Systems 05/2013; DOI:10.1007/978-3-642-36461-7_10
[Show abstract][Hide abstract] ABSTRACT: Background
The integration of empirical data in computational frameworks designed to model the spread of infectious diseases poses a number of challenges that are becoming more pressing with the increasing availability of high-resolution information on human mobility and contacts. This deluge of data has the potential to revolutionize the computational efforts aimed at simulating scenarios, designing containment strategies, and evaluating outcomes. However, the integration of highly detailed data sources yields models that are less transparent and general in their applicability. Hence, given a specific disease model, it is crucial to assess which representations of the raw data work best to inform the model, striking a balance between simplicity and detail.
We consider high-resolution data on the face-to-face interactions of individuals in a pediatric hospital ward, obtained by using wearable proximity sensors. We simulate the spread of a disease in this community by using an SEIR model on top of different mathematical representations of the empirical contact patterns. At the most detailed level, we take into account all contacts between individuals and their exact timing and order. Then, we build a hierarchy of coarse-grained representations of the contact patterns that preserve only partially the temporal and structural information available in the data. We compare the dynamics of the SEIR model across these representations.
We show that a contact matrix that only contains average contact durations between role classes fails to reproduce the size of the epidemic obtained using the high-resolution contact data and also fails to identify the most at-risk classes. We introduce a contact matrix of probability distributions that takes into account the heterogeneity of contact durations between (and within) classes of individuals, and we show that, in the case study presented, this representation yields a good approximation of the epidemic spreading properties obtained by using the high-resolution data.
Our results mark a first step towards the definition of synopses of high-resolution dynamic contact networks, providing a compact representation of contact patterns that can correctly inform computational models designed to discover risk groups and evaluate containment policies. We show in a typical case of a structured population that this novel kind of representation can preserve in simulation quantitative features of the epidemics that are crucial for their study and management.