Visa originalbild , Öppnas i nytt fönster/flik


Interactive Cognitive Architecture for Signal Processing

iCASP is a four years project funded by the Swedish Research Council (Vetenskapsrådet). The project is coordinated by Umeå University Hospital (Assoc. Prof. Urban Wiklund). Prof. Evgeny Osipov leads the LTU's part of the project. The projects funds the research of PhD student Tekn. Lic. Denis Kleyko.

The primary purpose of the project is to improve the diagnosis of autonomous nervous system  (ANS) dysregulation during clinical examinations and treatment. Specifically, the practical goal of the project is to improve the accuracy of the automatic detection and prediction of abnormalities in the cardiovascular autonomic regulation in real-time. The ANS is a subtle system, which makes sure that we are breathing and our heart keeps on pumping, where heart rate and blood pressure should be kept within safe limits. However, sometimes the regulation of the cardiovascular system does not work, or even breaks down. This dysfunction, which is observable by detecting abnormalities in cardiovascular signals, could be due to different diseases or damages in nerves, blood vessels, different organs, or in the brain itself. Automatic detection of abnormalities in the cardiovascular system has been a topic for extensive research during the past decades. However, on-line detection of changes in cardiovascular signals, such as acute treatment effects and severe cardiovascular events during diagnostic tests, is still a challenge and an extremely important problem to solve.

The novelty of the proposed project stems from the application of diverse machine learning techniques and a mathematical framework for biologically-inspired associative symbolic reasoning to reveal patterns when the number of variables are large (as after cardiovascular signal processing). Moreover, we take the challenge to do all analyses in real-time. The framework for associative symbolic reasoning is based on a novel methodology developed by the applicants that recently showed promising results for fault detection inside a human-made plant. We have asked ourselves—will this also work for fault detection inside of a human? This project will be one step towards answering this question.