Cognitive Cardiac Assessment Using Low-Cost Electrocardiogram Acquisition System

Abstract

Information and communication technologies are revolutionizing cardiac monitoring. Particularly, different Internet of Things (IoT) devices are gaining popularity, although basic cognitive tools that rely on electrocardiograms (ECGs) are still uncommon. Here, an ECG acquisition system for cognitive load analysis has been developed based on an Arduino-based, low-cost device for signal processing, recording, analysis, and classification. The system used network components such a cloud server for storage and related functions. By comparing the recorded signals to the reference professional medical device, the quality of the signals was confirmed. The Stroop test was used in the experiment to measure cognitive load in healthy subjects. The cognitive test caused, in most cases, characteristic changes in the structure of a large deviation multifractal spectrum. Thus, a new classification model based on multifractal total variations was presented for cognitive load assessment based on an ECG. The proposed cosine kNN (k nearest neighbors) approach yielded high accuracy results of above 90% using five-fold cross-validation, which were compared to other methods. It applied a relatively small number of features, including the Shannon entropy and the total variations.