Development of EDR Algorithm from ECG Signal Collected by an Ultra Low-Power Single-Lead ECG Device (wearable ECG)

The increasing demand of patients, undergoing chronic conditions, who wish to remain at home instead of being in a hospital and also the increasing requirement of  Homecare monitoring for elderly people, have resulted in a high request of wearable medical devices. Also, prolonged patient monitoring throughout daily activities has become a very critical goal. Low power consumption is essential in continuously monitoring of vital signs and MEC Medicine has developed EDR algorithms to extract respiratory rates out of ECG signals collected by an ultra low-power, single-lead ECG device. This device allows the patient to move with no constraint around an area, city or country. One of the other advantages is that we can exploit it to extract the respiratory signal during sleep and also intensive activities. Therefore, there is no longer need for cumbersome devices to record the respiratory activities. Our project in MEC Medicine investigated the feasibility and reliability of respiratory signals obtained from the forgoing device. The fusion of information from different sources is used throughout in developing algorithm, producing more accurate respiratory rate than that obtained using one source. To derive the respiratory rate from ECG, two approaches were followed in this study; one was based on variations in R peaks in QRS complex and the other was based on the effect of respiration activity on the ECG signal (i.e. Respiratory Sinus Arrhythmia (RSA)). In addition, to distinguish between valid and invalid respiratory signals, a confidence indicator was introduced. In fact, this confidence indicator sought information in the spectral domains of respiratory signals to find the number of dominant and non-dominant peaks as an index for dissociating the valid and invalid signals. Confidence indicator accomplishes this process by taking advantage of k-means clustering to distinguish dominant and non-dominant harmonics. The objectives of these approaches were to determine the relative contribution of each method, and to come up with novel methods. The results showed that there was a large reduction in mean error and standard deviation after applying data fusion technique as well as confidence indicator. The goal of this project was to evaluate the feasibility of obtaining respiration signals from single lead ECG recordings in different groups of subjects with advanced signal processing techniques. Images below, illustrate the process of project done by MEC Medicine Co.