A new bioinformatics system improves medical diagnosis
Researchers at the Universidad Politécnica de Madrid’s Facultad de Informática have developed an algorithm that improves the diagnosis and prognosis of many diseases, efficiently combining traditional clinical data with genetic data gathered using DNA microarray technology. The system uses DNA microarray technology to combine clinical and genetic diagnosis.
The algorithm, called CliDaPa (Clinical Data Partitioning), uses histological and clinical data and pharmacological treatments to partition patients by means of a tree representation for a particular disease (called clinical tree), used to cluster patients according to similar behaviour. It then uses data mining techniques to analyse each patient partition with the associated genetic information.
The method has been successfully validated on breast and lung cancer and on medulloblastomas (a type of brain tumour). Resources of Madrid’s Supercomputing and Visualization Centre (CeSViMa), one of the country’s major intensive computing infrastructures, were used to execute CliDaPa.
Compared with different studies reported in the scientific literature and traditional analysis techniques, the results show that CliDaPa offers a significant improvement on earlier results.
The research was developed as part of Santiago González Tortosa’s PhD thesis. The thesis was co-supervised by Victor Robles, of the Universidad Politécnica de Madrid’s Facultad de Informática, and Fazel Famili, of the National Research Council of Canada.
New approach
CliDaPa is a different method of DNA microarray analysis that aims to generate a model representing different patient behaviours (gathered from clinical data). These behaviours will then be examined separately and specifically by means of data mining. New learning methods were also proposed in the course of the research.
All this research has been published in international journals and congresses of repute. CliDaPa was also presented at the 8th UPM Business Start-up Competition and won an honourable mention.
Promising algorithm
A prominent biomedical application within the Cajal Blue Brain project is the comparative study of clinical data with factors sourced from complementary tests looking into the diagnosis and evolution of neurodegenerative diseases.
The research team is continuing research in the oncology domain in partnership with the Hospital de la Paz, as well as exploring new alternatives for pharmacogenomics data representation through the use of visualization, applying novel virtual reality techniques. The goal is to give medical experts advice on the behaviour of a specific disease based on different patient profiles. The use of this visual technique, combined with the expert’s previous experience, will help to diagnose and treat a disease more effectively and faster than current techniques are able to.
