Grants
Mobile phone data for epidemic risk
Mobile phones are nowadays one of the most pervasive technologies, reaching urban and rural populations across all socio-economic spectra. They allow to collect large-scale data providing extremely rich information on individual behavior. Every time individuals use mobile phones by making a call or sending a message, they produce a Call Detail Record (CDR). CDRs contain several attributes of the mobile phone activities including the caller and callee Ids, the type of the activity (incoming or outgoing call/SMS), the duration and the place and time when the mobile phone activity occurred. Such data provide the possibility to extract the social and mobility behaviors of the users. In fact, it can be used to extract the social interaction network based on social communications amongst individuals, or to reconstruct the individual trajectories by interpolating the displacements between any two consecutive activities. These reliable and detailed information are largely used for building spatially explicit epidemic models that take into account the contribution of human mobility to disease transmission.
In our lab, we use mobile phone data to inform non-Markovian models for disease spatial transmission.
In our lab, we use mobile phone data to inform non-Markovian models for disease spatial transmission.
We mainly have two lines of research. A theoretical one, where we assess how much of this high-resolution is needed for epidemic metapopulation models, depending on the disease and epidemic context. And an applied research direction, focusing on the role of mobility in the HIV epidemic in sub-Saharan Africa, and the interplay between mobility and social dimensions on the 2014 West Africa Ebola virus epidemic.
Epidemics on temporal networks
|
Temporal - or dynamic - networks are characterized by an evolving contact structure: links are created and destroyed as time passes. In recent years a lot of research has focused on temporal networks, as they have become an indispensable tool for studying human contacts (social, sexual..), transportation networks, and so on. Not only taking into account the temporal dimension of the network provides a deeper understanding of its underlying dynamics, but it is often a crucial factor to include in epidemic modeling. The interplay between the network evolution time scale and the disease time scale has been shown to influence conditions, speed and routes of epidemic spread. In particular the epidemic threshold, i.e. the critical value of the transmissibility above which the epidemic is able to invade the network, has been analytically computed so far only in specific settings and case studies.
|
|
Animal infectious diseases
Animal diseases represent an important threat to human health, since the emergence of human diseases is dominated by zoonotic pathogens. We work on infectious diseases affecting livestock and also wild animals, focusing on characterizing the conditions for the pathogen spread and maintenance in the host population.
A large amount of data is available for livestock movements, as European countries monitor farmed animals at the individual level and on a daily basis. Such monitoring efforts have led to a unique opportunity of studying animal movements in a comprehensive way, characterizing their behavior in time and space, and identifying patterns that may become relevant for the spread of a potential disease in the livestock population.
In particular, the coexistence of stationary heterogeneous behaviors at the system level with strong temporal fluctuations at the microscopic level limits our understanding of the disease spread through movements, and of its prevention, prediction, and control, because of:
(i) the strong dependence of the spreading pattern on the initial conditions, both geographical and temporal, and
(ii) the lack of meaningful definitions of nodes’ importance, given the observed large temporal fluctuations of measures of centrality based on static structural properties.
Novel methods of investigation, simulation and analysis need to be developed to face these challenges.
A large amount of data is available for livestock movements, as European countries monitor farmed animals at the individual level and on a daily basis. Such monitoring efforts have led to a unique opportunity of studying animal movements in a comprehensive way, characterizing their behavior in time and space, and identifying patterns that may become relevant for the spread of a potential disease in the livestock population.
In particular, the coexistence of stationary heterogeneous behaviors at the system level with strong temporal fluctuations at the microscopic level limits our understanding of the disease spread through movements, and of its prevention, prediction, and control, because of:
(i) the strong dependence of the spreading pattern on the initial conditions, both geographical and temporal, and
(ii) the lack of meaningful definitions of nodes’ importance, given the observed large temporal fluctuations of measures of centrality based on static structural properties.
Novel methods of investigation, simulation and analysis need to be developed to face these challenges.
Digital surveillance
vous etes grippe? faites-vous connaitre!
Determining the number of cases in an epidemic is fundamental to properly evaluate several disease features of high relevance for public health policies such as mortality, morbidity or hospitalization rates. Disease surveillance is key to control and response planning, providing valuable information to devise and apply public health interventions in time. The recent H1N1 influenza pandemic provided a paramount example of the crucial importance of surveillance, highlighting the need to improve surveillance capacities worldwide and real-time response against emerging diseases.
Influenza activity is currently monitored using a combination of clinical and virological information collected through a network of sentinel medical practitioners (GPs) and a network of reference laboratories. The French experience of the Reseau Sentinelles counts more than 20 years of GP surveillance of communicable diseases in France. Developed in 1984 by INSERM in collaboration with the Ministry of Health, and taking advantage of computer systems and the Internet, it now includes about 1200 voluntary sentinel GPs who remotely enter reports on 12 conditions on a weekly basis. Results are accessible in form of plots, maps and reports on a publicly available website and are used by experts in the field for research purposes, but also by the general public and the media to get information on the current evolution. This huge longitudinal catalogue of disease surveillance data have provided invaluable information to conduct epidemiological research addressing a wide range of topics, from feeding modeling studies to the analysis of the geotemporal spreading patterns, the assessment of interventions on influenza morbidity, the estimation of relevant epidemiological quantities for seasonal influenza and emerging influenza pandemics.
Next to the traditional surveillance system, innovative Information Communication Technologies (ICT) are spurring the development of new mechanisms for information retrieval, and now start to allow the prompt gathering of large amount of data enabling for the first time the real-time analysis of a pandemic in a very detailed way. In particular, the collaborative aspect of the Web2.0 is now pervading a large set of systems including health-related ones. Influenzanet.org is an online project for disease surveillance that emerged in this context and that exploits collaborative aspects, pervasiveness of connection, and ICT to collect surveillance data at high spatial and temporal scales on a large European region. The system consists of a fast and flexible method to monitor in real time the temporal and geographic evolution of influenza activity in several European countries through the participation of volunteers in the population. Citizens living in those countries can freely and voluntarily participate, answering a weekly survey of questions for the assessment of influenza-like-illness (ILI) based on reported symptoms. The information provided by the users is treated according to the national and European privacy regulations, and is processed in real time and visualized in form of maps and plots, accessible directly on the website, and publicly available.
Determining the number of cases in an epidemic is fundamental to properly evaluate several disease features of high relevance for public health policies such as mortality, morbidity or hospitalization rates. Disease surveillance is key to control and response planning, providing valuable information to devise and apply public health interventions in time. The recent H1N1 influenza pandemic provided a paramount example of the crucial importance of surveillance, highlighting the need to improve surveillance capacities worldwide and real-time response against emerging diseases.
Influenza activity is currently monitored using a combination of clinical and virological information collected through a network of sentinel medical practitioners (GPs) and a network of reference laboratories. The French experience of the Reseau Sentinelles counts more than 20 years of GP surveillance of communicable diseases in France. Developed in 1984 by INSERM in collaboration with the Ministry of Health, and taking advantage of computer systems and the Internet, it now includes about 1200 voluntary sentinel GPs who remotely enter reports on 12 conditions on a weekly basis. Results are accessible in form of plots, maps and reports on a publicly available website and are used by experts in the field for research purposes, but also by the general public and the media to get information on the current evolution. This huge longitudinal catalogue of disease surveillance data have provided invaluable information to conduct epidemiological research addressing a wide range of topics, from feeding modeling studies to the analysis of the geotemporal spreading patterns, the assessment of interventions on influenza morbidity, the estimation of relevant epidemiological quantities for seasonal influenza and emerging influenza pandemics.
Next to the traditional surveillance system, innovative Information Communication Technologies (ICT) are spurring the development of new mechanisms for information retrieval, and now start to allow the prompt gathering of large amount of data enabling for the first time the real-time analysis of a pandemic in a very detailed way. In particular, the collaborative aspect of the Web2.0 is now pervading a large set of systems including health-related ones. Influenzanet.org is an online project for disease surveillance that emerged in this context and that exploits collaborative aspects, pervasiveness of connection, and ICT to collect surveillance data at high spatial and temporal scales on a large European region. The system consists of a fast and flexible method to monitor in real time the temporal and geographic evolution of influenza activity in several European countries through the participation of volunteers in the population. Citizens living in those countries can freely and voluntarily participate, answering a weekly survey of questions for the assessment of influenza-like-illness (ILI) based on reported symptoms. The information provided by the users is treated according to the national and European privacy regulations, and is processed in real time and visualized in form of maps and plots, accessible directly on the website, and publicly available.
The system was originally launched in the Netherlands and Belgium in the 2003/2004 influenza season (www.degrotegriepmeting.nl) mainly as an experiment of scientific communication. It had a great success, reaching about 30K active participants and was then implemented in Portugal from the 2005/2006 season (www.gripenet.pt) and in Italy in the 2008/2009 season (www.influweb.it). Starting 2009, it became part of the European Commission Integrating Project EPIWORK, aiming at the extension of the system to the entire Europe. New countries have joined the project since then, including the UK ( www.flusurvey.co.uk) and Sweden (www.influensakoll.se), and soon also Germany, Austria, Switzerland will follow.
Vittoria Colizza collaborated with the Computational Epidemiology lab @ISI Foundation for the development, set-up and scientific coordination of the italian system, influweb, that was launched in 2008 and today counts about 5000 users.
On January 25, 2012, in collaboration with the Reseau Sentinelles @INSERM & UPMC and with the InVS, we launched the online influenza system in France, under the name of grippenet.fr.
Do you live in france? help us monitor the influenza in the country, register on grippenet.fr and let us know how you feel.
Do you live in france? help us monitor the influenza in the country, register on grippenet.fr and let us know how you feel.
Emerging epidemics
(Re-)emerging pathogens, such as MERS-CoV and Ebola, are source of large medical burden and economic costs. The early stage of the outbreak presents biological and epidemiological uncertainties that often hinder a prompt medical and health care response thus increasing the harm associated to the event. Improving our capabilities to assess the epidemic situation and its associated risk represents a fundamental scientific challenge with a clear public health impact.
We combine complex system approaches, extensive data integration, computational programing and epidemiological statistics to study emerging pathogen events in real time in order to provide understandings, risk assessment and projections. Crucial modelling framework within this effort is GLEAM, the global epidemic and mobility model, that combines real-world data on population and human mobility with ad hoc infectious disease dynamics to simulate epidemic events and provide scenarios analysis. A publicly available software based on GLEAM, the GLEAMviz Simulator, can be downloaded from the project webpage.
We combine complex system approaches, extensive data integration, computational programing and epidemiological statistics to study emerging pathogen events in real time in order to provide understandings, risk assessment and projections. Crucial modelling framework within this effort is GLEAM, the global epidemic and mobility model, that combines real-world data on population and human mobility with ad hoc infectious disease dynamics to simulate epidemic events and provide scenarios analysis. A publicly available software based on GLEAM, the GLEAMviz Simulator, can be downloaded from the project webpage.
Hosts heterogeneities and epidemic spread
Human behavior displays many degrees of heterogeneity. Contact patterns among individuals, frequency and duration of travels, topology of transportation systems and fluxes of travelers, as well as many other aspects related to human interactions and mobility are shown to be broadly distributed and to largely fluctuate among different classes of individuals. Great part of epidemic modeling studies are based on simplifying assumptions typically relying on homogeneous approximations regarding some of the aspects mentioned above. Our research work in this area seeks to go beyond these simplifications aimed at reaching an understanding of the impact of the observed heterogeneities on the propagation of an infectious disease.Paragraph. Clicca qui per modificare.
|
At the spatial level, human mobility is a key ingredient of epidemic propagation. In the last few years, thanks to the availability of massive data records and computational resources to analyze them, a large effort has been devoted to the investigation of the fundamental laws of human mobility at all scales. Leveraging on this increasing body of knowledge we focus on a number of different aspects of human mobility (cultural/geographic/socio-economic aspects) and their role on epidemic spreading. Furthermore, we integrate large mobility datasets in mathematical and computational models to increase their realism and allow for explicit simulations of entire populations up to the scale of single individuals.
|
|
|
At a different scale with respect to human mobility, we are also interested in analyzing and modeling human f2f interactions. Nowadays it is becoming increasingly easy to collect high-resolution data through sensing devices. The heterogeneities, correlations and temporal variability of these data call for the design of new modeling schemes for assessing the impact of the observed interaction patterns on infectious disease transmission and outbreak risk.
|
DATAREDUX - Big data reduction for predictive computational modelling
|
3DataRedux project focuses on developing radically new methods for the reduction of the complexity of large networked datasets to feed effective and realistic data-driven models of spreading phenomena. Many rich datasets on actions and interactions of individuals have recently become available, commonly encoded as networked systems, arising from heterogeneous sources with details at different scales and resolutions, and potentially containing geographical and temporal information as well as metadata. These outstanding sources of information and knowledge fuel a wide spectrum of data-driven numerical simulations of dynamical processes. Data alone, however, even in huge amounts, do not easily transform into knowledge or predictive models. The rich and diverse information they contain raises crucial challenges concerning their analysis, representation and interpretation, the extraction of meaningful structures, and their integration into data-driven models. In this context, DataRedux puts forward an innovative framework to reduce networked data complexity while preserving its richness, by working at intermediate scales (“mesoscales”). Our objective is to reach a fundamental breakthrough in the theoretical understanding and representation of rich and complex networked datasets for use in predictive data-driven models for decision making and actionable insights.
Duration: 4 years, 2020-2023 Partners: CNRS (Centre national de la recherche scientifique), ENS-Lyon (École normale supérieure de Lyon), INSERM Our role: WP leader Website: soon available |
RISKFLOW - Uncovering HIV risk flow networks to improve current approaches for controlling HIV epidemics in sub-Saharan Africa
|
Achieving global elimination of HIV will require intensive efforts to decrease incidence and increase treatment coverage in sub-Saharan Africa (SSA), where most HIV-infected individuals live. and transmission is predominantly heterosexual. As a consequence of heterosexual transmission, the prevalence of HIV is high in the general population. Therefore, HIV epidemics in SSA are generalized rather than concentrated in high-risk groups, as is the case in resource-rich countries. Highly mobile populations in SSA is expected to substantially increase the difficulty to eliminate HIV. We propose here to develop a fundamentally new approach for controlling HIV epidemics in SSA. With the use of mobile phone data collected countrywide over one year, including 9 billion communications from ~1.2 million unique SIM cards, we assess to what extent the mobility of individuals imposes a spatial structure on the generalized HIV epidemic in Namibia and how this structure can be used to design new health policy tools.
Duration: 2 years, 2019-2021 Partners: INSERM, UCLA (University of California Los Angeles) Our role: Coordinator Website: soon available |
CDR4Ebola - Socio-behavioral response to 2014 West Africa Ebola virus epidemic measured from mobile phone data
|
The current context of urbanization, high mobility, massive trade, and climate change make our modern world increasingly vulnerable to biological invasions. Pathogens of pandemic potential continue to emerge at an alarming rate. During the last twenty years alone, we have seen the emergence of SARS, MERS, and Zika, as well as the re-emergence of Ebola in humans. Urbanization, expanding transportation systems and increasing mobility volumes are all major drivers for spatial transmission, creating the perfect storm of conditions for infectious disease emergence and propagation. At the same time, opportunities for mixing and conditions for traveling may be altered by fear of the ongoing epidemic, or because of the implementation of travel restrictions in certain periods. Relying on high-resolution data in space and time from mobile phones, this project will assess the socio-behavioral changes of individuals during the 2014 Ebola epidemic in Guinea and how such adaptation impacted the observed spread.
Duration: 3 years, 2019-2022 Partners: INSERM, INRIA (French research institute for digital sciences) Our role: Coordinator Website: soon available |
SPHINX - Spread of Pathogens on Healthcare Institutions Networks
|
The project proposes a global approach to better understand and control the spread of HAI integreating various scales (hospital wards, hospitals and healthcare facilities network,community) by developing a multi-scale computational framework. The way human populations are structured within and outside hospitals, the transfers of patients between wards and between hospitals, as well as the drug exposure in different settings, are all scales intrinsically interacting with the biological scale of pathogen transmission. Most importantly, all these scales are bound to have interrelated impacts on the intervention strategies aiming at controlling HAI spread. Mathematical models, developed jointly with epidemiological investigations, are a powerful tool to better understand the mechanisms of HAI spread in healthcare settings. However, previously published modelling studies were typically limited to describing HAI spread at a single scale, independently of the others. This scale separation challenges our ability to get a global understanding of this phenomenon. In addition, each scale is characterized by network contacts – contacts between individuals due to proximity, contacts between healthcare institutions due to transfers of patients, contacts between patches of populations due to mobility and community/hospital interactions, which are seldom taken into account in models.
Duration: 4 years, 2018-2021 Partners: CNAM (Conservatoire national des Arts et métiers), Institut Pasteur, EHESP (Ecole des hautes études en santé publique), AP-HP (Assistance Publique Hôpitaux de Paris), INSERM Our role: WP leader Website: https://sites.cnam.fr/sphinx/ |