БГHome | Team Members | Aims | Scientific publications and presentations
The project presents fundamental research in the fields of mathematics and computer science, biomedical engineering and medicine, related to solving a major societal challenge to improve quality of life – health, by developing innovative and intelligent technologies in favor of the diagnosis of cardiovascular disease. The objective of the project proposal is to design, train, optimize and test on clinical data new software algorithms with deep machine learning. The aim is to expand the application of artificial intelligence in emergency medicine, screening and ambulatory programs through more accurate and effective computer support for early diagnosis of cardiac rhythm pathologies by means of automatic interpretation of the electrocardiogram (ECG). The focus is on deep neural networks (DNNs), which are hypothesized to be able to self-learn on large datasets to extract unexplored information from characteristic details of patterns unique to a particular type of pathology at deep hierarchical levels, and as a result to achieve much higher accuracy than conventional (shallow) machine learning techniques. The main tasks are combined in the following work packages (WP):
- WP2: Application of DNNs for analysis of life-threatening cardiac arrhythmias in automatic external defibrillators.
- WP3: Application of DNNs for pre-processing of ECG signals.
- WP4: Design of a clinical database and new methods for accurate, automated diagnosis of various arrhythmias in ECGs recorded by long-term (24 h) Holter-ECG monitoring.
- WP5: Application of DNNs for precise, automated measurement of diagnostically useful ECG parameters and classification of arrhythmias in multichannel electrocardiographic recordings.
