Mainly relying on a multidisciplinary approach, the ERIAS research group brings together computer science researchers with public health medical doctors specialized in medical informatics with the aim at providing methods and tools to support the activities of data collection, data integration, handling incertitude, implicitness and temporality of healthcare data while dealing with data provenance. The main objective is to turn raw data into high quality data for enabling secondary use of healthcare data.
With the rapid development of Information Technology (IT), the availability of numeric data in various domains is increasing. Yet, it is possible to obtain from them meaningful facts that can lead to new knowledge. In the public health domain in particular, great is the opportunity to speed up and facilitate research activities. Moreover, the availability of additional data enables the enlargement of the scope of data to exploit (e.g., a better representability of the studied population or a supplementary data dimension available for a population). Thereby, facilitating secondary use of different kinds of data is a promising approach from the public health perspective.
However, in addition to being available, data have to be adapted for secondary use purpose. Automating the tasks required for such adaptations is raising new challenges. Indeed, IT-based approaches need to take into account the specificities of health-related data for being successful. The major issues raised by such data are: (i) the inherent incertitude that they may hold but also (ii) their temporal facet as well as (iii) the implicit meaning they convey, which often depends on common knowledge shared by physicians that may not be explicitly expressed. These specificities combined with the complexity of conducted research in the meantime require specific approaches. They can benefit from the long tradition of health professionals to classify things for describing the biomedical domain although the rapid evolution of knowledge in this field constitutes an additional challenge. Overcoming these issues needs a pluridisciplinary approach, involving both medical and computer science skills.
Data of Interest
Electronic health records
Electronic health records (EHRs) are recognized as a major source of information for research purpose. Even when research data are acquired by hand, EHRs are often used. Indeed, a possible workflow is to manually review patient’s EHRs in order to enter data within databases specifically structured for research purpose. As a result, enabling direct reuse of EHRs in order to automatically feed research databases is an essential issue and is the subject of numerous studies for several years. Secondary use of EHRs is not a straightforward task. Indeed, when reusing data in EHRs, multiple challenges are to be addressed including semantic and syntactic data normalization, phenotype identification and data quality evaluation.
Web-based data
The Web constitutes nowadays a new channel of information. On the one hand, the context of modern healthcare practice requires to access to recent findings as reported in the scientific literature. On the other hand, data that are being generated in the context of social networks may be a useful resource which contains real-time data and may complement those from databases provided by official bodies or fed by EHRs. However, although valuable, these data are unstructured and expressed in natural language, sometimes in a poor, informal or degraded vocabulary (online health related forums for instance). With the objective to make use of such data for being able to address use-cases for which no other data can be exploited, ERIAS provides methods and tools in order to tackle Web-based data. For instance, in the pharmacovigilance domain, the identification of adverse-drug events as well as drug misuses may benefit from the exploitation of such data.
Main research axis: Data and Knowledge integration for public health
With Electronic Health Records and Web based data, scientific bottlenecks are related to the heterogeneity of resources, the volume of data and the fact that a large part of such data are unstructured and convey implicit meaning. Indeed, those data exhibit heterogeneities at different levels (structural, content, etc.) and are scattered across multiple sources, which are themselves described using their own syntax and semantics. For a homogeneous and comprehensive view of these data, it is first necessary to extract and annotate them according to common notions using formal languages, and then reconcile them, when necessary, through an integration process. In addition, another important challenge is to develop advanced approaches to efficiently access to relevant information. These three essential steps are closely interlinked and are being addressed by the ERIAS research group under the common umbrella of data and knowledge integration for public health.
The three topic addressed under this umbrella are :
- Information Extraction and annotation,
- Knowledge reconciliation
- Enabling efficient access to relevant information.
Main publications
- Ayllón-Benítez, Aarón, Fleur Mougin, Jesualdo Tomás Fernández-Breis, Manuel Quesada Martinez, Patricia Thebault, et Romain Bourqui. 2017. « Deciphering gene sets annotations with ontology based visualization ». In Proceedings of 21st International Conference Information Visualisation (IV), 2017. London, United Kingdom. https://hal.archives-ouvertes.fr/hal-01524572.
- Jouhet, Vianney, Fleur Mougin, Bérénice Bréchat, et Frantz Thiessard. 2017. « Building a Model for Disease Classification Integration in Oncology, an Approach Based on the National Cancer Institute Thesaurus ». Journal of Biomedical Semantics 8 (1): 6. https://doi.org/10.1186/s13326-017-0114-4.
- Mougin, Fleur, David Auber, Romain Bourqui, Gayo Diallo, Isabelle Dutour, Vianney Jouhet, Frantz Thiessard, Rodolphe Thiébaut, et Patricia Thébault. 2018. « Visualizing Omics and Clinical Data: Which Challenges for Dealing with Their Variety? » Methods 132 (janvier): 3‑18. https://doi.org/10.1016/j.ymeth.2017.08.012.
- Nikiema, Jean Noël, Vianney Jouhet, et Fleur Mougin. 2017. « Integrating Cancer Diagnosis Terminologies Based on Logical Definitions of SNOMED CT Concepts ». Journal of Biomedical Informatics 74 (octobre): 46–58. https://doi.org/10.1016/j.jbi.2017.08.013.
- Petit-Monéger, Aurélie, Frantz Thiessard, Pernelle Noize, Driss Berdaï, Vianney Jouhet, Florence Saillour-Glénisson, Louis-Rachid Salmi, PACHA research group, et others. 2017. « Definition of indicators of the appropriateness of oral anticoagulant prescriptions in hospitalized adults: Literature review and consensus (PACHA study) ». Archives of Cardiovascular Diseases.
- Dramé, Khadim, Fleur Mougin, et Gayo Diallo. 2016. « Large Scale Biomedical Texts Classification: A KNN and an ESA-Based Approaches ». Journal of Biomedical Semantics 7 (juin): 40. https://doi.org/10.1186/s13326-016-0073-1.
- Diallo, Gayo. 2014. « An Effective Method of Large Scale Ontology Matching ». Journal of Biomedical Semantics 5 (1): 44. https://doi.org/10.1186/2041-1480-5-44.
- Avillach, Paul, Jean-Charles Dufour, Gayo Diallo, Francesco Salvo, Michel Joubert, Frantz Thiessard, Fleur Mougin, et al. 2013. « Design and Validation of an Automated Method to Detect Known Adverse Drug Reactions in MEDLINE: A Contribution from the EU-ADR Project ». Journal of the American Medical Informatics Association: JAMIA 20 (3): 446‑52. https://doi.org/10.1136/amiajnl-2012-001083.
- Thiessard, Frantz, Fleur Mougin, Gayo Diallo, Vianney Jouhet, Sébastien Cossin, Nicolas Garcelon, Boris Campillo, et al. 2012. « RAVEL: retrieval and visualization in ELectronic health records ». Studies in health technology and informatics 180: 194‑98.
Details
Centre de recherche INSERM U1219
Université de Bordeaux – ISPED
146 rue Léo-Saignat
33076 BORDEAUX cedex
Tél : +33 (0)5 57 57 15 04
Fax : +33 (0)5 57 57 45 33
Informations
News
- 17 Feb 2020
Conference March 10, 2020 – Sustainable IT and some of its challenges
This event has been organised by the A.C.E group – Action Climate Environment leading by Research and Teaching…
