ECG Classification Algorithm Based on STDP and R-STDP Neural Networks for Real-Time Monitoring on Ultra Low-Power Personal Wearable Devices

Abstract

This paper presents a novel ECG classification algorithm for real-time cardiac monitoring on ultra low-power wearable devices. The proposed solution is based on spiking neural networks which are the third generation of neural networks. In specific, we employ spike-timing dependent plasticity (STDP), and reward-modulated STDP (R-STDP), in which the model weights are trained according to the timings of spike signals, and reward or punishment signals. Experiments show that the proposed solution is suitable for real-time operation, achieves comparable accuracy with respect to previous methods, and more importantly, its energy consumption is significantly smaller than previous neural network based solutions.

A review of wearable and unobtrusive sensing technologies for chronic disease management

“…Moreover, these approaches require no ECG segmentation, or only QRS detection (a simple task as mentioned earlier), and are more robust to ECG noise [ 140 ], [ 141 ]. Deep learning-based methods have also been carefully optimized and can be deployed into mainstream wearable devices for real-time heartbeats screening [ 142 ], [ 143 ].…”

A Neural Network-Based ECG Classification Processor With Exploitation of Heartbeat Similarity

“…Two widely used statistical measures of F1 score and G score adopted in [7] are listed to synthetically estimate the Sen and Ppr. An overall performance comparison between the proposed algorithm and other advanced patient-specific methods [7], [12], [13], [21] are shown in Table 3. Our classification performance of V beats is comparable to existing methods, while the classification performance of S beats is more outstanding than others.…”

“…The Sen, F1 and G of the S beats classification are the best among all. It is worth mentioning that the [21] proposes an efficient low-power design for neural network based ECG classification. However, it only achieves four-class (N, V, F, and Q) heartbeat recognition, and the classification accuracy of V can still be improved further.…”

A Hybrid Heartbeats Classification Approach Based on Marine Predators Algorithm and Convolution Neural Networks

“…Various studies have been conducted using computer-aided detection (CADe) to precisely forecast arrhythmia in order to enhance diagnostic efficiency. [6], [7], [8]. Thus, This work aims to early alerts for unusual disorders may be given by creating a computational framework focused on machine learning that rapidly, correctly, and consistently diagnose cardiac arrhythmia such that qualified doctors can provide appropriate care.…”