En poursuivant votre navigation sur ce site, vous acceptez le dépôt de cookies dans votre navigateur. (En savoir plus)

H/F - PhD Thesis: An interdisciplinary contribution toward Covid-19 pandemic space time evolution assessment

This offer is available in the following languages:
Français - Anglais

Date Limite Candidature : lundi 30 mai 2022

Assurez-vous que votre profil candidat soit correctement renseigné avant de postuler. Les informations de votre profil complètent celles associées à chaque candidature. Afin d’augmenter votre visibilité sur notre Portail Emploi et ainsi permettre aux recruteurs de consulter votre profil candidat, vous avez la possibilité de déposer votre CV dans notre CVThèque en un clic !

General information

Reference : UMR5672-PATABR-003
Workplace : LYON 07
Date of publication : Monday, May 9, 2022
Scientific Responsible name : Patrice Abry
Type of Contract : PhD Student contract / Thesis offer
Contract Period : 36 months
Start date of the thesis : 1 September 2022
Proportion of work : Full time
Remuneration : 2 135,00 € gross monthly

Description of the thesis topic

key-words: Covid-19, interdisciplinarity, optimization, estimation, stochastic sampling. Research Unit: CNRS, ENS Lyon - Physics department & Rhoˆne-Alpes Complex
System Institute (ixxi.fr) - Lyon, France Responsables : Patrice Abry, patrice.abry@ens-lyon.fr,
web : http://perso.ens-lyon.fr/patrice.abry
Context: The Covid-19 pandemic has constituted a major stress-test for all societies, revealing weaknesses and inducing multiple challenges very different in nature. The worlds of research and academia were also hit in at least two fundamental aspects: i) the practice of interdisciplinary research; ii) the relations sciences / society.
The health crisis has indeed generated numerous scientific questions involving all dis- ciplines: biology, epidemiology, but also statistical processing of information, political science, anthropology, philosophy, etc. Its analysis therefore requires a fundamentally multidisciplinary approach. However, in the academic world, two rigorous, well-founded, rational and high quality logics are being deployed, but they are struggling to fertilize each other: that of the specialists in numbers and modeling, and that of the specialists in social sciences and critical thinking. This observation calls for a paradigm shift: rather than making disciplines converge towards the same object based on their specific advances, the goal is to put the object (the health crisis) at the center and to stimulate interdisciplinary exchanges based on the questions it raises.
Moreover, the health emergency, like other crises (the climate emergency, for example), directly exposes scientific statements to citizens (the general public) and decision makers (politicians). This has resulted in a pendulum swing between an initial strong demand for scientific support for pandemic control strategies and a distrust of scientific discourse and analysis. In this field, our hypothesis is that science must improve its informational function towards citizens and decision-makers. This program, torn between the antagonistic needs to quickly produce knowledge on an unknown field and to share it in a debate outside the perimeter of the places of knowledge, calls for the continuation, renewal and reinvention of the modes of interaction between science and society.
In this context, researchers from Ens de Lyon, organized around the Institute of Com- plex Systems (IXXI), bringing together researchers from the Physics Laboratory (SEE) and the Triangle Laboratory (SHS), spontaneously gathered to work together to measure, mapp and model the spatio-temporal evolution of the reproduction rate of the pandemic. We have already developed a procedure that allows, from public data (often of limited quality), to obtain an estimate of the reproduction number R(t) [4, 3], regular in time (piecewise linear) and space (e.g. piecewise constant across departments), and thus realistic and usable in practice. This estimate is formulated as a convex (non-smooth) minimization problem [2], which allows for non-linear filtering (smoothing, denoising) of the data [1]. We then auto- mated and made available daily updates of these spatio-temporal evolutions, for more than two hundred countries and for the departments of metropolitan France, through interactive and animated maps, for both informed and wide audiences, willing to access robust and documented scientific information: http://barthes.enssib.fr/coronavirus/cartes/ Rmonde/ and http://barthes.enssib.fr/coronavirus/cartes/RFrance/ (cf. Fig; 1).
This work has aroused great interest. Nevertheless, it suffers from important method- ological limitations (sensitivity to low data quality, outliers, non-stationarity of the pan- demic, adaptation to a given territory); it would benefit from integrating richer datasets and other parameters, especially social ones; and from being enriched with communication tools in order to facilitate the debate.
Goals and Research program: Research tracks are organized along two main direc- tions, combining and gathering in an interdisciplinary frame statistical signal processing and applied math with humanities, social sciences and cartography.
On one hand, we will develop an integrated set of statistical information processing tools to measure, predict, visualize, understand, and interpret the spatio-temporal evolution of the Covid-19 pandemic. This implies to address several statistical signal processing issues:
• Instead of a three-step procedure (preprocessing of outliers, estimation, regulariza- tion), we will perform these three processes jointly and optimally. This implies to be able to build several functional proposals that take into account the presence of outliers, that are conceptually relevant and concretely minimizable. And therefore to make these functionals robust to the non-stationarity of the pandemic, for example by exploring an adaptive update, driven by the data rather than supervised by an expert: ”hyperparameters” that regulate a) the importance given to the outliers, b) the degree of regularization (the equivalent bandwidth of the filter).
• The estimates would benefit from being accompanied by confidence intervals. For this, a theoretical investigation on the potential use stochastic sampling in a pro- cedure intrinsically deterministic is necessary. It will be based on the mastery of trajectory sampling techniques (bootstrap, MCMC. . .).
• Current estimates are based on the number of new daily infections when richer in- formation is available (hospitalizations, deaths, resuscitations). Therefore, we will build, from the epidemiological literature, a multivariate model keeping the qualities of the current univariate, rich enough to preserve the phenomenology of a pandemic and simple enough to be usable.
• The graph of territories (departments), so far built on the naive notion of shared borders, will be rethought to take into account the complexity of social interactions (train distances, regional hospitals, resorts, demographic disparities, etc.).
On other hand, the outcomes of these tools will be used to address societal stakes raised by the pandemic. Notably, the research work will contribute to the production of animated, configurable and interactive map permitting to visualize jointly the space-time evolution of the intensity of the pandemics and to compare them between countries. Such maps will be both used to work on comparative and quantitative assessments of the impacts of the various sanitary politics defined by Public Health Authorities (lockdown, corfew, vaccinations, ...) and to favor direct transfer of information from scientists to citizens. This may be done by quantifying the usability of the developed maps for a general (non necessarily scientific) audience. This may also mean to design materials to enrich maps with hypertextual information: evolution of the R as a function of time, display of a panel of relevant meta-data, themselves configurable to understand and analyze this evolution (and according to expectations: socio-demography, modeling, etc.), moreover adapted to the considered territories.
2

Figure 1: Space-time evolution fo the reproduction number R(t). Animated and in- teractive maps available at http://barthes.enssib.fr/coronavirus/cartes/Rmonde/ and http://barthes.enssib.fr/coronavirus/cartes/RFrance/.
Skills: The candidate should have a strong taste for interdisciplinary work and show motivation both for theoretical statistical developments, real data analysis and for inter- actions with cartography and humanities. Python, Matlab, SVG skills would be a plus.
Application: Please send CV and motivation letter to patrice.abry@ens-lyon.fr.
References
[1] Guichard E. et al. Abry P. Spatial and temporal regularization to estimate covid-19 reproduction number R (t): Promoting piecewise smoothness via convex optimization. PlosOne, Aug. 2020. URL: https://journals.plos.org/plosone/article?id=10. 1371/journal.pone.0237901.
[2] Heinz H Bauschke, Regina S Burachik, Patrick L Combettes, Veit Elser, D Russell Luke, and Henry Wolkowicz. Fixed-point algorithms for inverse problems in science and engineering, volume 49. Springer Science & Business Media, 2011.
[3] F. Brauer, C. Castillo-Chavez, and Z. Feng. Mathematical models in epidemiology. Springer, New York, 2019. URL: https://www.math.purdue.edu/~fengz/pub/book_ contents.pdf.
[4] A. Cori, N. M. Ferguson, C. Fraser, and S. Cauchemez. A new framework and soft- ware to estimate time-varying reproduction numbers during epidemics. American Jour- nal of Epidemiology, 178(9):1505–1512, 2013. URL: https://academic.oup.com/aje/ article/178/9/1505/89262.

Work Context

Ecole Normale Supérieure de Lyon,
Physics Department Institut Rhône-alpes complex system Institut
team : Signaux, Systèmes et Physique (SiSyPhe)

Constraints and risks

N/A

Additional Information

Skills: The candidate should have a strong taste for interdisciplinary work and show motivation both for theoretical statistical developments, real data analysis and for inter- actions with cartography and humanities. Python, Matlab, SVG skills would be a plus.
Application: Please provide a CV, a motivation letter and recommendation letter(s)

We talk about it on Twitter!