Offre en lien avec l’Action/le Réseau : – — –/– — –
Laboratoire/Entreprise : CRIStAL labs (computer science Lab – https://www.c
Durée : 2 years
Contact : clarisse.dhaenens@univ-lille.fr
Date limite de publication : 2020-12-31
Contexte :
Applications are invited for one 2-year postdoctoral fellow to begin on January 1, 2021. The position will be co-supervised by Pr. Clarisse Dhaenens, computer Science, Pr. Vincent Sobanski, internal medicine and Pr. Grégoire Ficheur, medical informatics.
The postdoc is part of a research cluster from I-Site ULNE (http://www.isite-ulne.fr/index.php/en/home/).
She/He will be hosted in the CRIStAL labs (computer science Lab – https://www.cristal.univ-lille.fr/ – Univ. Lille, CNRS, Centrale Lille)
and is financed by Lille metropole (MEL – https://www.lillemetropole.fr/)
She/He will also work actively with the INCLUDE team at the Lille University Hospital (health data warehouse project – https://include-project.chru-lille.fr/en/the-team/), the METRICS team (ULR 2694, Lille University Hospital), and the INFINITE team (INSERM U1286, Lille University Hospital – http://lille-inflammation-research.org/en/workpackages/861-wp3-en).
Sujet :
Project objectives:
The objective of this project is to use a biclustering approach to combine two different and complementary data sources in order to better characterize endotypes of chronic inflammatory diseases:
• Build a “word embedding” type representation based on BERT modeling (Bidirectional Encoder Representations from Transformers, Devlin 2018) from textual concepts present in medical letters from a computerized patient records database
• Filter a set of characteristics unrelated to the type of diagnosis that may be found among the interests of patients
• Build clusters by the biclustering method, based on (i) the identified signs, (ii) on the previously constructed embedding and / or (iii) on the variables from a local complete database
• Compare the clusters obtained using these different approaches, in particular with regard to the concordance on the grouping of patients as well as on the clinical relevance of the subgroups identified.
Profil du candidat :
The applicant should have a PhD in Computer Science.
Formation et compétences requises :
The ideal candidate would have expertise in one or more of the following areas:
Operations research, Combinatorial Optimization, Data Mining (clustering, bi-clustering), optimization approaches, textual analysis…
An interest or an experience about health domain will be appreciated.
The position is open to candidates of any nationality and selection will be based upon the candidate’s research record and potential.
Adresse d’emploi :
She/He will be hosted in the CRIStAL labs (computer science Lab – https://www.cristal.univ-lille.fr/ – Univ. Lille, CNRS, Centrale Lille)