A particular focus of our research within data science is on ensemble methods, i.e., techniques for generating sets of models that collectively form predictions by voting, and on methods for generating interpretable models, e.g., rule learning.

data mining(2).jpg

Another focus is on text mining, in particular efficient and resource lean methods using language technology for very large text sets. The research also focuses on semantic analysis, e.g., negation, speculation and temporality, in order to be able to extract situation-specific, accurate and relevant information from texts.

One main application area for the research at the department is health care analytics, which aims for providing efficient and effective decision support for health care and pharmaceutical research.

The project High-Performance Data Mining for Drug Effect Detection is supported by the Foundation for Strategic Research with 19 MSEK during 2012-2016. The main goal of the project is to develop techniques and tools to support decision making and discovery of drug effects by analysing patient records, drug registries, case safety reports and chemical compound data in the form of both structured and unstructured (free text) data. The project will contribute with novel approaches to data mining and clinical text mining and develop a platform for large-scale analysis of massive, heterogeneous and continuously growing data sets.

The research group has collaborated for several years with computational chemists in the pharmaceutical industry. This has resulted in new techniques and tools for building predictive models from observed biological activities, e.g., toxicity, of chemical compounds, which are currently being used in the industry.

Another application area that we are involved in is modeling of component wear in heavy trucks using data mining and to provide decision support for optimizing heavy truck fleet utilization. We participate in the project Integrated Dynamic Prognostic Maintenance Support (IRIS), which is lead by Scania AB, and supported by 11.6 MSEK from Swedish Governmental Agency for Innovation Systems (VINNOVA) during 2012-2017.