Web Service for Automatic Detection of R-Peaks in Electrocardiograms

Abstract

The analysis of the electrocardiogram (ECG) is an essential tool for the diagnosis of cardiovascular diseases. Sometimes, it is necessary to carry out long-term recordings, even up to 24 hours, which must be carefully evaluated by specialists in order to generate an accurate diagnosis. In the case of extensive records, this task becomes monotonous, tedious and susceptible to errors, highlighting the need to implement Artificial Intelligence (AI) to automate this process, which would be a valuable support for health professionals. However, it is important to note that most devices used for ECG recording are low capacity, making them incompatible for running AI algorithms directly on them.  An implementation of a web service for the identification of R-peaks in electrocardiograms is presented. This service performs the identification by means of a convolutional neural network, which after being trained reached a sensitivity of 0.99658 and a specificity of 0.99655. In this work, the main objective is to present a basic proposal to generate services that help to perform robust data analysis with high processing power consumption for low-resource devices such as microprocessors.