Ontologie, Données et Informatique médicale

Mise à jour (2019/06/07): Les diapos des présentations sont maintenant disponibles sur le site – voir ci-dessous dans le programme.

Une journée scientifique en anglais sur le thème “Ontologie, Données et Informatique médicale” est organisée à l’IRIT le lundi 27 mai 2019, avec le soutien du domaine d’application stratégique “Santé, autonomie, bien-être” et de l’action stratégique “Calcul, masses de données, IA”.

Deux experts en utilisation d’ontologies pour traiter les données médicales viendront présenter leurs travaux :

– Prof. Mark Musen (université de Stanford)

– Prof. Jean-François Ethier (université de Sherbrooke)

Des membres de l’IRIT (voir programme ci-dessous) viendront également présenter leurs travaux ou projets. Des présentations en fin de journée élargiront la discussion à l’utilisation d’ontologies dans d’autres domaines que le médical.

La journée scientifique est ouverte à tous.

Quand ? Lundi 27 mai, 9h – 17h

Où ? Auditorium Herbrand, Institut de Recherche en Informatique de Toulouse (IRIT), rez-de-chaussée (Campus de l’université Paul Sabatier, Métro Université Paul Sabatier)

Langue de présentation : AnglaisInscription : Si vous pensez assister à la journée, il est préférable de s’inscrire sur https://www.irit.fr/survey/index.php/876654 pour être certain d’avoir une place (l’inscription au buffet du repas de midi est en revanche désormais terminée).

Orateurs principaux :

Mark Musen est professeur d’informatique biomédicale à l’université de Stanford, où il dirige le “Stanford Center for Biomedical Informatics Research” et le “WHO Collaborating Center for Classification, Terminology, and Standards”. Sa recherche concerne les technologies sémantiques (ontologies, métadonnées) à l’appui des systèmes intelligents et systèmes d’aide à la décision biomédicale. Son groupe a développé Protégé, l’outil le plus utilisé à l’heure actuelle pour le développement et la gestion des terminologies et ontologies.

Mark Musen présentera le projet CEDAR.

Jean-François Ethier est clinicien-chercheur et professeur agrégé au département de médecine à l’Université de Sherbrooke, où il co-dirige le Groupe de recherche interdisciplinaire en informatique de la santé (GRIIS). Sa recherche s’intéresse aux systèmes de santé apprenants à base d’ontologie, qui développent des méthodes d’interopérabilité sémantique entre les systèmes de santé et de recherche afin de favoriser la ré-utilisation des données cliniques, génomiques, administratives et issues des appareils personnels connectés (quantified self).

Jean-François Ethier présentera le projet PARS3.

Programme de la journée :

(programme sous réserve de modifications mineures)

9h – 10h : Mark Musen (Stanford University): Semantic Technology for Open Science: Creating an Ecosystem for FAIR Data

With the explosion of interest in open science, the past few years have seen considerable discussion about making scientific data “FAIR”—findable, accessible, interoperable, and reusable.  The problem is that most scientific datasets are not FAIR.  When left to their own devices, scientists do an absolutely terrible job creating the metadata that describe the experimental datasets that make their way in online repositories.  The lack of standardization makes it extremely difficult for other investigators to locate relevant datasets, to re-analyze them, and to integrate those datasets with other  data.  The Center for Expanded Data Annotation and Retrieval (CEDAR) has the goal of enhancing the authoring of experimental metadata to make online datasets more useful to the scientific community.  CEDAR illustrates the importance of semantic technology to driving open science. It also demonstrates a means for simplifying access to scientific data sets and enhancing the reuse of the data to drive new discoveries.

10h-10h30 : Lynda Tamine-Lechani (équipe IRIS) : Representation learning of documents driven by external knowledge resources

An intensive recent research work investigated the combined use of hand-curated knowledge resources and corpus-driven resources to learn effective text representations. The overall learning process could be run by online revising the learning objective or by offline refining an original learned representation. In this talk I will focus on the design of comparable offline vs. online knowledge-enhanced document representation learning models and the comparison of their effectiveness using a set of standard information retrieval (IR) and natural language processing (NLP) downstream tasks. Through intensive experiments carried out using datasets and knowledge resources from the general domain and the medical domain, we will provide lessons to support future frameworks of knowledge-enhanced document representation learning.

10h30-11h : Coffee break

11h-11h30 : Imen Megdiche-Bousarsar (équipe SIG) : Aligning ontologies in the context of large size and number

To address the problem of aligning large biomedical ontologies, we propose an innovative system combining machine-learning techniques and “divide and conquer” approach to match these resources  with high quality results. The second part of the speech will be devoted to present the problem of aligning large number of ontologies simultaneously known as holistic matching.  We address the bottlenecks and our proposed solution.

11h30-12h : Cassia Trojahn (équipe MELODI) : Complex alignments for better covering ontology heterogeneities

Simple ontology alignments, largely studied in the literature, link a single entity of a source ontology to a single entity of a target ontology. A limitation of these alignments is their lack of expressiveness which can be overcome by complex alignments. In this presentation we introduce complex alignments and present the complex matching approach that has been developed in the MELODI team.

12h-13h30 : Lunch buffet (inscriptions closed)

13h30 – 14h30 : Jean-François Ethier (GRIIS, Université de Sherbrooke): PARS3, an infrastructure to support data access activities in context of learning health systems

Learning health systems (LHS) are now recognized as important frameworks to structure activities with the aim of improving citizens’ well being. While a LHS requires close coupling of care delivery, research and knowledge transfer processes, challenges remain, especially in the case of LHS involving more than one organisations. These larger scale LHS are necessary to have a more complete understanding of an individual since relevant data is be fragmented across various organisations. This is a requirements for multiple types of projects like evaluating care trajectories. The session will review briefly the LHS framework and challenges it aims at solving in terms of data access. It will also present an example of a LHS platform supporting data access developed by the Quebec SPOR Support Unit, PARS3, as well as how this architecture will be deployed in various scales: local, provincial and national levels.

14h30 – 15h : Adrien Barton (équipe MELODI) : An ontological analysis of drug prescriptions

The ambiguities and overspecificities of prescription semantics along with their lack of standardization limit data interoperability and are potential sources of error. An ontology of drug prescriptions could help overcome such difficulties and ultimately reduce adverse drug events. I will present the Prescription of Drugs Ontology (PDRO), a reference ontology used in the PARS3 project (see above), founded on the Open Biomedical Ontology Foundry (OBO Foundry) realist principles and built on the upper ontology Basic Formal Ontology (BFO). It defines a key component of prescriptions, a drug administration specification, as an action specification that directs a process of drug administration. I will present a classification of the kind of instructions that can be found in a drug prescriptions and an analysis of their internal structure.

15h – 15h30 : Rémi Bastide (équipe SIG) : Semantic components for IoT healtcare applications

e-Health is an application domain of choice for IoT applications. We present a software architecture designed for such applications, based on the exchange of semantic information between loosely coupled components. This approach, coupled with a proper design process allow allow for linking domain-independent ontologies such as SSN (Semantic Sensor Ontology) to domain-specific health ontologies.

15h30-16h : Coffee break

The last block will be devoted to ontological projects out of the medical domain:

16h – 16h30 : Nathalie Aussenac-Gilles (équipe MELODI) : The ontology projects of the team MELODI

16h30 – 17h : Amina Annane (équipe MELODI) : Business process ontology

Any industrial company has its own business processes, which is a number of related tasks that have to be executed to reach well-defined goals. In order to analyze, improve, simulate and automate these processes, it is essential to represent them in a formal way. The activity of representing business processes is known as Business Process Modelling (BPM); it is an active area of research that attracts more and more attention these last years with the emergence of Industry 4.0. Semantic Web technologies, especially ontologies, are promising technologies to advance BPM and realize the Industry 4.0 vision. We present a novel ontology for business process representation, that we called BBO. We developed BBO using existing meta-models and ontologies. In particular, BPMN 2.0, the state-of-the-art meta-model for business process representation. This work is part of the AVIREX project in collaboration with large industrial companies: Thales Alenia Space and Continental.

Soutiens :

Domaine d’application stratégique “Santé, autonomie, bien-être” (IRIT)

Action stratégique “Calcul, masses de données, IA” (IRIT)

Organisation : Adrien Barton, Nathalie Souf, Nathalie Aussenac-Gilles