R & D Activities

The research directions of the Computational Medicine Laboratory are based on the realisation that

a future where genetic profiling or patient stratification based upon genetic variance is routine is not that difficult to imagine. Diagnosis based on genotypic and integrated phenotypic data (i.e. clinical genomics) will result in more effective treatments earlier, extending the life of the population and improving the quality of life. Readily available, integrated patient data will help to identify patients at risk for adverse drug reactions, improve clinical trials and drug discovery and tailor individualized treatment for a variety of diseases, and
the challenge for the knowledge-enabled healthcare enterprises of the future will be the successful, flexible integration of distributed information and knowledge across distributed application components to support and continuously improve business processes as well as the timely and effective delivery of appropriate information to all authorized users (researchers, health professionals and citizens at large) of such an enterprise.

The Laboratory is focusing on various computational aspects of biomedical informatics, such as (a) ontology based integration and analysis of genetic and medical information for health applications; (b) Grid-based approaches to demanding molecular-biomedical applications; (c) analysis, simulation and modeling of complex biomedical processes, and (d) design and development of novel and prototypical DM/KDD methods, techniques, algorithms, tools and systems through four independent but highly correlated and interdisciplinary activities.

These are:

eHealth Activity. Our R&D priorities in this area are focusing around the issues related to the creation of an integrated electronic health record (I-EHR) for every citizen, by addressing key challenges related to the provision of a framework for the integration of a diverse set of heterogeneous and distributed information sources into what appears to be a uniform collection of data and knowledge, so as to increase the availability of previously inaccessible information.
We are also focusing on the requirement of transforming such an I-EHR from a passive into an active record, so as to support a patient-centered, clinically driven healthcare system. Towards this end Computational Medicine Laboratory is focusing on issues related to (a) linking the I-EHR to external knowledge sources such as clinical guidelines/protocols and genetic information; and (b) the development of predictive models for diseases/treatments. The objective being to improve medical knowledge through the elicitation of currently unknown correlations ("non-hypothesis based medicine") between an individuals' clinical history and the risk of developing new pathologies and between medical treatments and unwanted side effects.
Additional R&D efforts are related to the important requirement for significantly improving the care delivery processes through coordination of multiple human and other resources that are spread over multiple organisations in an enterprise as well as across multiple enterprises. To this end, Computational Medicine Laboratory's R&D efforts focus on the development of new, innovative ambient intelligence service platforms for automatic, context sensitive offering and contracting of eHealth and mobile Health (mHealth) services across heterogeneous networks
Data Mining and Knowledge Discovery from Databases Activity. The R&D efforts of the DM/KDD activity are directed towards four directions aiming to expose the utility of data-mining in the respective disciplines and application areas.

These are:

Design and development of novel and prototypical DM/KDD methods, techniques, algorithms, tools and systems (e.g., MICSL - Multiple Iterative Constraint Satisfaction based Learning);
Intelligent analysis approaches, based on DM/KDD techniques, for the recognition of genes (promoters) in DNA sequences; Intelligent analysis of microarray/gene-expression data for the discovery of molecular/gene markers based on feature-selection approaches, and the discovery of families of co-regulated genes (metagenes) based on clustering approaches (e.g., MineGene integrated microarray data analysis tool); Design and development of integrated clinico-genomic decision making methodologies
Design of methodologies and development of algorithms, tools and systems for mining distributed and heterogeneous clinical data sources. The aim is to add intelligent capabilities into the integrated electronic healthcare record (I-EHR) towards internet-based epidemiology (e.g., HealthObs - a system for the discovery of interesting clinical associations from distributed and heterogeneous clinical information systems, based on Association Rule Mining operations) and
Design and development of techniques, algorithms, systems and tools for: the automated categorisation of Web-documents into users´ interests, and the automated construction concept hierarchies.

Medical and Molecular Imaging Activity. The scope of the activity is to provide the necessary technology and tools for optimizing biomedical information for Clinicians and Biologists. The Bioimaging activity is involved in several research directions, encompassing different aims: (a) The Medical and Molecular Imaging research, aims to develop methods for multi-modality image alignment and fusion that enables temporal studies (e.g. cell trafficking, fluorophore detection etc.) to be carried out more efficiently, (b) The Gene expression imaging research aims to develop sophisticated image analysis algorithms for signal normalization, and (c) The Nanoimaging research is a new field with very little participation of Greece internationally. The aim is to provide significant contributions towards the development of automated image analysis software for the nanomanipulation of biological structures (e.g. DNA)
Bioinformatics Activity. The Bioinformatics activity develops novel mathematical and computational tools that can be used to solve complex, data intensive biological problems. Methods developed and used include symbolic and sub-symbolic machine learning algorithms, as well as global optimization methods. These developments are compared and combined with existing tools. Specific objectives include:

Prediction of novel precursors microRNA (miRNA) from genomic sequences using novel methods for the calculation of RNA secondary (2D) / tertiary (3D) structure
Prediction of mature miRNA from precursor miRNA
Computational approach for prediction of miRNA targets
Graph-theoretical analysis and visualization of regulatory networks and cellular simulation
Combined gene and miRNA expression analysis
Support for the development of novel nanotechnology Lab-on-chip devices for diagnostic miRNA-gene expression profiles, as well as
Proteome annotation for cellular localisation.

The Laboratory is also contributing substantially to the activities of the Biomedical Informatics Working Group of ERCIM.