A Novel Approach for Robust Detection of Heart Beats in Multimodal Data using Neural Networks and Boosted Trees

Abstract

This work describes a novel approach designed for Physionet 2014 Challenge, Robust Detection of Heart Beats in Multimodal Data [5]. The objective here is to detect the location of R peaks from QRS complex of an electrocardiogram (ECG) excerpt. Robust detection of heart beats in a noisy ECG signal is an extremely difficult task. To overcome the challenge in such situations, besides ECG, blood pressure (BP) signal is also recorded at the same time; hence the idea here is that, if a segment of one of the signals is noisy, the peaks in that segment can be better estimated by peaks found in the corresponding segment of other signal, if good.The approach uses Machine Learning (ML) methods to identify locations of R-peaks in a given segment of ECG or BP signal. Peaks from both ECG and BP signal are found separately using a novel feature representation and subsequent ML approaches that renders R peaks in the signal, easier to be detected, by a simple windowing technique. Individually detected peaks, from both ECG and BP are further analyzed in chunks of equal short time periods, and the best result of the two is chosen in final peak prediction based on variance comparison techniques. The performance of system on the training dataset [11] provided in the competition is 99.95%. The performance on test datasets which are hidden for phase I, phase II and phase III of the competition respectively are 93.27%, 90.28% and 89.74%. The submission resulted in 1 st place in all three phases of the competition.

ECG Monitoring Systems: Review, Architecture, Processes, and Key Challenges

“…Supporting processes, such as training and modeling, were highlighted to support machine learning and neural network techniques [68,73]. Data cleansing was also defined as a discrete process in [71].…”

“…AI methods and neural networks are typically very useful in providing ECG signal interpretation. In the following, we depict a few examples from the literature of different techniques used for ECG signal processing [68,[98][99][100][101]. Recent research in the literature adopts Neural Network (NN) and decision trees for the diagnosis of different cardiac diseases, the assessment of cardiac health conditions, the detection of chronic problems, sleeping issues including apnea, and mood and emotion recognition [55].…”

“…Numerous monitoring systems define modeling, learning and prediction [75]. Some research proposals define a distinct training process [35,68] deep learning (ANN) process [83,87], machine learning (ML) process [86], and ML streaming process [13].…”

Multiple Physiological Signals Fusion Techniques for Improving Heartbeat Detection: A Review

“…The use of ECG and BP signals has also been extensively studied due to their direct relationship with cardiac activity [58,59,60,61,62,63,64,65], from which other approaches that integrate up to 3 additional signals have also been presented: SV [66], EEG [67,68], and EOG [69,70] signals; EOG and EMG signals [71]; SV and EOG signals [72]; EEG, EOG, and EMG signals [73]; and SV, EEG, and PPG signals [74].…”

Signal-Level Fusion With Convolutional Neural Networks for Capacitively Coupled ECG in the Car

“…In the course of the challenge, several algorithms were developed and tested on a hidden dataset containing various multimodal clinical signals. As a follow-up to the original challenge, an algorithm for beat detection was developed that uses individual fully connected neural networks (FCNs), which are then fused using gradient boosted trees [10]. Similarly, the work presented in [11] uses a CNN for ECG annotation, while the majority of recent works that use CNNs are aimed at direct classification, for example of cardiac arrhythmias.…”