CARDIALLY developed new techniques for real-time measurements for the electrocardiogram (ECG) applications and beyond. New techniques of real-time signal processing including machine-learning methods have been developed and tested for analysis of ECG signals. In the framework of the project, we demonstrated how mutual information can be used as a metric for ECG analysis. Recent advances in development of portable ECG devices that required real time processing make efficiency and simplicity of implementation of measures an especially important feature. In the framework of the CARDIALLY, we compared performance of different distance measures and use them as identifier of abnormalities between different cycles of ECG. We demonstrated that the Jaccard distance (based on mutual information) provides a simple and efficient way to identify abnormality in heart beating (here we studied effects of arrhythmia and fibrillation), robust to translation invariance and noise. We showed that it can be used for identification of DM changes of ECG heart beats even in the presence of strong noise and when the baseline changes between ECG heart beats. This is important for portable devices. The simplicity of the method makes it a promising candidate to be used as part of classification/clustering algorithm or in portable ECG-analyzing devices as straightforward and simple identifier of abnormal behaviour of ECG.
As to abnormal behaviour of ECG we emphasize that, in the framework of the CARDIALLY, we considered the implementation of Algorithms to predict shock outcome based on ventricular fibrillation waveform. Indeed, VF is a medical emergency of enormous proportions and it is one of the first causes of sudden death in a large range of population's age. Different algorithms have been constructed, characterized, tested and evaluated exploiting the observational prospective study, done during the CARDIALLY project, which includes 260 patients with out-of-hospital cardiac arrest treated by the emergency medical services in Brescia, Italy, between 2006 and 2009. We demonstrated that Algorithms can be potentially useful tools to effectively optimize defibrillation strategy, thus immediate defibrillation versus cardiopulmonary resuscitation.
The main achievement of the CARDIALLY is a demonstration of the prototype device fabricated by DIASENS. All the knowledge transfer targets, that required synergetic efforts of all participating teams and cross-training of staff, have been successfully achieved. DIASENS applied for permission to organize pilot clinical testing of the prototype device and the request for approval is in the process of consideration. The draft protocol for the pilot clinical testing was designed in collaboration with the medical experts and working cardiologists from the Medical School of University of Belgrade (
http://www.mfub.bg.ac.rs/eng/home/(s’ouvre dans une nouvelle fenêtre)).
During the project period DIASENS in collaboration with the partner institutions completed hardware of the experimental multi sensor device and appropriate software for control the measuring process and data acquisition and advanced numerical methods for cost efficient data processing. The system started from the basic idea reached the technology readiness level TR4. It is fully operational and ready to proceed with the planned data acquisition in the hospital environment. The successful realization of the experimental multi sensor device is the result of the gained knowledge of the DIASENS researchers from the partner institutions through the completed extensive secondments plan on design, fabrication and encapsulation of the fiber optical sensors, and advanced algorithms for signal processing and data analyses.
