uncompressed-src Engineering health information. Methods and tools for the acquisition, modeling and formalization of knowledge for e-health
- Design methods and tools for the development and evaluation of terminological and ontological resources in health;
- Develop Web services for accessing the strings of treatments on exploited resources and strive to set of micro-formats for accessing the data in the context of the \" Web of data \" ;
- Scale representation of data models to take into account the specificities of the field of health (data and multimodal knowledge, multi-scale, temporal, temporal, spatial 2D and 3D and geotagged) and semantically annotate these data with the resources built to enable their integration;
- Participate in the development of the new international repositories of Health Informatics (information, terminologies and ontologies in health models), valuing the French work in international cooperation and ensure the alignment of French solutions on the most advanced solutions in the world.
The issues specific to this research axis are illustrated by the following examples :
- Searching for information in the patient record, data warehouses, literature, guides of good practices or on the Internet is today essential to decision-making and the re-use of information. The "correct" interpretation of this information at the international level is a major issue in our societies. The semantics associated with the information must be developed, standardized, standardized and comply with the scalability of the information.
- Recent developments of new technological approaches in health product of huge masses of very heterogeneous data. Semantic integration of these data is a major challenge for the understanding of the mechanisms that regulate our health. The semantics associated with the information (data and vocabularies models) must be published and aligned with the standards to be shared. In this great mass of heterogeneous data, it happens that certain information be hidden due to the size of the space where they are located (finding a needle in a haystack), or because of the implicit information related to their representation. The issue of semantic integration is to allow linking of data more accurately and to extract new knowledge.
- In the area of health, there are many different terminologies because their creation and use were made at different times for different reasons, and it is impossible to remove them without sacrificing research information or information exchange systems that use. Reconcile these terminologies corresponds to a matching effort which is particularly heavy. The challenge is to produce semantic resources harmonised in the field so that researchers can compare and aggregate data for the purpose of multiple analyses. Moreover, the semantics associated with the information must be accessible so modular, adaptable and interoperable.
To meet these challenges, we propose 1) to develop approaches to formalize the semantics of a domain and make the applications that can integrate and process structured and encoded data within operational information systems and 2) help (including within standardization bodies) the sharing of models of health informatics by providing methods and tools development and collaborative maintenance of these models. Methodological locks are much related to the complexity of the medical field that seeks health professionals extremely diverse - clinicians, pharmacists, biologists, nutritionists etc. - and the patients themselves.
A field of this research is dedicated to the construction of ontologies and the development of information systems to health operator knowledge formalized within these ontologies. These are the formal representations of semantics more adapted to the computer operations of knowledge [Gruber 95]. Ontology is the notions of a domain by organizing them in the form of a hierarchy of concepts, i.e. a taxonomy [Guarino 92; Bachimont 00; Smith 01]. On the one hand, the ontology to develop, standardize, to standardize the semantics associated with the information allowing the user to use terminology that is familiar to describe this information. On the other hand, it allows the semantic integration of heterogeneous data. For example, renal artery stenosis, a mutation on a gene or excessive salt consumption are data related to the concept of hypertension that can be aggregated in a reasoning if the link is explained in the model of knowledge that describes the concept of hypertension (i.e. l'ontology).