Abstract
Internet of Things (IoT) has transformed healthcare systems in a huge manner since it allows doctors keep a check on patients’ health in real time, especially those with cardiac problems. Electrocardiographic (ECG) data are particularly important for detecting cardiovascular issues early on. Electrocardiograms are sensitive to noise and distortions, which can make it hard to undertake an analysis that is both quick and accurate. The tools we have now for looking at ECGs either have too many steps or aren’t accurate enough. This is because the ways these systems get features are either fixed or not very deep. These limits make it tougher to keep an eye on things in real time, which slows down speedy diagnosis and makes it harder to utilize on IoT devices that don’t have a lot of resources. The results of this study show that it could be a good idea to use a Hybrid Adaptive Feature Extraction (HAFE) method in an IoT architecture to handle ECG inputs. The HAFE additionally has statistical analysis for reducing features, adaptive signal decomposition using empirical mode decomposition (EMD), and time-frequency localization with discrete wavelet transform (DWT). We employ a convolutional neural network (CNN) that is set up to work on the edge to sort these properties. The system can execute analytics in real time because it runs on a Raspberry Pi 3 computer and is backed up by the cloud. For instance, it was 98.6% accurate, 97.9% sensitive, and took 1.7 seconds to make a prediction.
Authors
S. Karthiga1, S. Sathish Kumar2
Thiagarajar College of Engineering, India1, Veltech Hightech Dr Rangarajan Dr Sakunthala Engineering College, India2
Keywords
ECG Signal, IoT Healthcare, Feature Extraction, Adaptive Hybrid Algorithm, Real-Time Monitoring