Project contact: jaak.vilo (at) stacc.ee

Biomedical data integration and mining addresses the fact that ever more electronic data is being gathered in the health system.Many health files are already electronic, laboratory data are being added and biomarker methods are being improved.While most of these processes are concentrated on gathering and presenting the data of individual patients, a higher objective is the statistical analysis of global trends in order to improve the entire health system and make it more information-based.We have launched four biomedical research projects that complement each other to pursue this objective.

Cohort of smokers for predicting COPD

The object of our research is Chronic Obstructive Pulmonary Disease (COPD), a complex but widespread and dangerous disease.The objective is to develop methods for an automatically identifying higher risk smokers from their early complaints, to monitor these patients, and to study potential genetic and other biomarkers using blood and other analyses.The medical side is managed by Professor Ruth Sepper from the Technomedicum of Tallinn Technical University.Data analysis requires data from the medical system to identify smokers, analyse their complaints and establish ontologies.

Data warehouse

Although a lot of medical information concentrates on individual patients’ complaints and medical data, the data of all patients are needed simultaneously for hospital management and for epidemiological and methodological research.The data warehouse project aims to develop solutions to collect data in a uniform and comparable form and to integrate them into a data warehouse, where global analyses can be carried out by hospital management and specialist doctors.For instance, research may look at patients with similar diagnoses by comparing two or more typical medical treatment methods or medicaments and identifying their advantages in different circumstances.

The data can also be used for forecasting trends and for better planning of resources.The data warehouse should allow ad hoc queries – questions that have not been asked before.The design of a data warehouse is complicated by the diversity of data, as different diseases may have very different and unusual parameters.The goal is to determine which services are highest priority for hospital management.

Ontologies and other semantic assets

Data analysis is generally possible if data can be classified.The aim of this project is to find the best types of classification, ontology and multiple choice for data.The issue is not about the form of classification or the description of the available information, but how it can be presented so that the doctors and nurses as end users could access the data in the shortest possible time.

Analysis of medical free texts

STACC News