NCFU mathematicians have developed an effective system to identify cardiovascular issues through ECG signals. The approach has proven more accurate than any other ones, and allows improving the quality of early detection of heart problems, which is of extreme importance in terms of dangerous disease prevention.

– We have developed a new architecture with an advanced recurrent neural network structure, - Pavel Lyakhov (project manager, Head of Dept for Mathematical Modeling; Head of Dept for Modular Computing and Artificial Intelligence, Regional Research & Training Mathematical Center, NCFU) noted. – To ensure more accurate interpretation of signals, we use preliminary digital data processing, which is done through digital filters, spectral analysis and some other methods. Given this, we have managed to reduce significantly various noises that distort the ECG signal. This resulted in a higher accuracy of data interpretation – up to 87.5 per cent. As far as the best foreign methods are concerned, this index reaches 79-83 per cent.

The research has been carried out with funding support from the Russian Foundation for Basic Research and Presidential Council for Grants.

More details concerning the project are available in a special issue of Advanced Information Processing Methods and Their Applications (Applied Sciences Journal). To learn more, please follow this link.