Medical technologies: Bioinformatics and biomedical signals

Where we are

Centre de Recerca en Enginyeria Biomèdica (CREB) UPC

How to reach us

Related websites

Researchers from the Biomedical Signals and Systems group offer solutions to companies, hospitals and institutions for the development of technologies applicable to the hospital setting, including their use in surgical instruments.

Scientific objectives

  1. To develop and validate a set of methods for the functional enrichment of Omics data based on diffusive algorithms.
  2. To carry out a descriptive statistical analysis of rare diseases in collaboration with the Innovation Department of the Sant Joan de Déu Research Institute (IRSJD) and within the framework of the European Share4Rare project.
  3. To publish new data analysis methods primarily relating to metabolomics and gene expression. To develop these methods, including enrichment analysis using the previous knowledge of existing annotation databases such as KEGG and Reactome. We will soon publish a new version of MAIT (the first version is available at Bioconductor), including an interactive on-line platform.
  4. To investigate new clinical risk indices based on non-invasive biosignals.

Area/Field of expertise

Our most outstanding accomplishment within our first line of research is our progress on functional enrichment approaches through diffusion-based techniques applicable to experiments with data from different omic technologies. Our results are published in the open source bioinformatics repository, Bioconductor, on diffusion methods in biology knowledge graph representation in KEGG (Bioconductor - FELLA) offering a methodology for functional enrichment in metabolomic experiments. We have also published standard diffusion algorithms (Bioconductor - diffuStats) as well as various data processing tools, especially in metabolomics (such as Bioconductor - MAIT, CRAN - eRah, CRAN - baitmet) or in genetic association (b2slab - MISS).

Regarding our second research line, we have specialised in multivariate and automatic learning techniques for the analysis of large volumes of clinical, biomedical and biochemical data. Of note is the analysis carried out of a cohort of 1,500 patients diagnosed with heart failure over a 15-year follow-up period; the paper has recently been accepted for publication by JACC: Journal of the American College of Cardiology (IF 16.8, in collaboration with the Germans Trias i Pujol University Hospital, Barcelona). We have also gained expertise, the technical and high-performance computing capacity in the use of Deep Learning algorithms. We could mention, for example, the use of this technology to construct predictive models of protein-compound interaction from recurrent neural network models (Long Short-Term Memory) based on amino acid sequences and SMILES codes, adjusted by TensorFlow (in collaboration with Mind the Byte S.L.).

In terms of signal analysis, we are particularly interested in cardiovascular research using morphology signal analysis in electrocardiography (ECG), RR interval analysis (for heart rate variability [HRV]) and depth of anaesthesia using electroencephalography (EEG) signals, amongst others. Our research goal will be to strengthen the research methods used in the B2SLab regarding the use of linear techniques, symbolic dynamics, information theory, multiscale complexity and multifractality.

Group members

Last Publications

More Publications


  • Multiview pattern recognition methods for data visualization, embedding and clustering
    Kanaan Izquierdo, Samir
  • Design and development of methods for paediatric tumours classification using DNA methylation profiles
    Llano Viles, Joshua
More Theses


See All news