sensors

Article
Convolutional Autoencoding and Gaussian Mixture Clustering
for Unsupervised Beat-to-Beat Heart Rate Estimation of
Electrocardiograms from Wearable Sensors
Jun Zhong 1,2 , Dong Hai 1 , Jiaxin Cheng 3 , Changzhe Jiao 3 , Shuiping Gou 3 , Yongfeng Liu 1, *, Hong Zhou 1
and Wenliang Zhu 1

1 Suzhou Institute of Biomedical Engineering and Technology, Chinese Academy of Sciences,

Suzhou 215163, China; [email protected] (J.Z.); [email protected] (D.H.); [email protected] (H.Z.);
[email protected] (W.Z.)
2 Suzhou Institute of Biomedical Engineering and Technology, University of Science and Technology of China,
Hefei 230026, China
3 Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education of China,
School of Artificial Intelligence, Xidian University, Xi’an 710071, China; [email protected] (J.C.);
[email protected] (C.J.); [email protected] (S.G.)
* Correspondence: [email protected]; Tel.: +86-512-6958-8055

Abstract: Heart rate is one of the most important diagnostic bases for cardiovascular disease. This
paper introduces a deep autoencoding strategy into feature extraction of electrocardiogram (ECG)





 signals, and proposes a beat-to-beat heart rate estimation method based on convolution autoencoding
and Gaussian mixture clustering. The high-level heartbeat features were first extracted in an unsuper-
Citation: Zhong, J.; Hai, D.; Cheng,
vised manner by training the convolutional autoencoder network, and then the adaptive Gaussian
J.; Jiao, C.; Gou, S.; Liu, Y.; Zhou, H.;
mixture clustering was applied to detect the heartbeat locations from the extracted features, and
Zhu, W. Convolutional Autoencoding
and Gaussian Mixture Clustering for
calculated the beat-to-beat heart rate. Compared with the existing heartbeat classification/detection
Unsupervised Beat-to-Beat Heart Rate methods, the proposed unsupervised feature learning and heartbeat clustering method does not rely
Estimation of Electrocardiograms from on accurate labeling of each heartbeat location, which could save a lot of time and effort in human
Wearable Sensors. Sensors 2021, 21, annotations. Experimental results demonstrate that the proposed method maintains better accuracy
7163. https://doi.org/10.3390/ and generalization ability compared with the existing ECG heart rate estimation methods and could
s21217163 be a robust long-time heart rate monitoring solution for wearable ECG devices.

Academic Editors: Yeon Sik Noh and Keywords: autoencoding; electrocardiogram; Gaussian mixture clustering; heart rate estimation
Toshiyo Tamura

Received: 11 July 2021

Accepted: 25 October 2021
1. Introduction
Published: 28 October 2021
Based on data from the National Health and Nutrition Examination Survey (NHANES),
Publisher’s Note: MDPI stays neutral
2013 to 2016, the prevalence of cardiovascular disease (CVD) in adults over 20 years of age
with regard to jurisdictional claims in
is 48.0% overall in the US (121.5 million in 2016) and increases with age in both males and
published maps and institutional affil- females. CVD prevalence, excluding hypertension, is 9.0% overall (24.3 million in 2016) [1].
iations. Cardiovascular disease has also become the number one disease, with high incidence
rates in the world [2]. Some researchers have proposed that it is necessary to carry out
real-time monitoring for patients with basic heart disease [3]. If real-time monitoring of
heart health can be carried out in daily life, the disease can be found as early as possible and
Copyright: © 2021 by the authors.
be treated in time, which is more conducive to improving the living quality of patients [4].
Licensee MDPI, Basel, Switzerland.
Heart rate is one of the most important ways to evaluate health, especially for patients
This article is an open access article
with cardiovascular disease. Heart rate variability, as an important indicator of health, is
distributed under the terms and helpful in the early diagnosis of diseases [5]. Therefore, developing robust long-term heart
conditions of the Creative Commons monitoring devices and beat-to-beat heart rate estimation methods are of great importance.
Attribution (CC BY) license (https:// An ECG is the most widely applied heart monitoring method in clinics. An ECG
creativecommons.org/licenses/by/ signal contains rich health information, reflecting vital signs. Through the analysis of
4.0/). an ECG signal, many indicators reflecting heart functions can be obtained, which is of

Sensors 2021, 21, 7163. https://doi.org/10.3390/s21217163 https://www.mdpi.com/journal/sensors

Sensors 2021, 21, 7163 2 of 18

great significance for subsequent diagnosis and treatment. ECG signals maintain the
advantages of stability and accuracy, which are ideal for heart health analysis. However,
a traditional ECG device requires electrodes to be attached to the patient’s chest, and
requires professional operating, which is inconvenient for daily operation. At present,
many wearable devices for heart rate measurements have been developed, such as chest
belts and vests [6–11]. These wearable devices are comfortable to wear; they are easy
for operation, at any time or for long-term monitoring. These wearable smart devices
record ECG, respiration, body temperature, and other related physiological parameters,
which are more conducive to the diagnosis of diseases. For patients, it can better achieve a
real-time analysis of the disease, with instant preventive treatment; for healthy people, it
can monitor and record the relevant heart condition changes during exercise, and can carry
out health monitoring on various indicators of the body in daily life. Figure 1 shows an
ECG embedded vest developed by the Suzhou Institute of Biomedical Engineering and
Technology, Chinese Academy of Sciences, which can collect human ECG signals in real
time. The vest records ECG signals through three embedded fabric ECG electrodes, which
can be comfortably worn in daily life.

Figure 1. A wearable ECG device that collects the ECG signals through three embedded fabric
electrodes in the vest.

For the signals collected by these wearable devices, because they do not directly
contact the skin, the signal quality of these electrodes is not as good as that measured
by the medical electrocardiograph. In addition to the baseline signal drift and hardware
frequency interference, subjects may inevitably have noise interference caused by sudden
moving or even moving in the long-term, which cast challenges to signal analysis from the
wearable devices.
Heart rate estimation from ECG data using supervised machine learning methods
typically require adequate amounts of training data, with accurate heartbeat annotations.
However, obtaining such volumes of labeled data could cost tremendous time and effort,
and is sometimes infeasible. For example, accurately labeling each cardiac cycle may require
high medical expertise and dedicated attention from the medical analyst. Ground truth
obtained from other intrusive measurements, such as a finger sensor, may bring intrinsic
inaccuracy to the ground truth when the signal quality is low, e.g., timing misalignment
Sensors 2021, 21, 7163 3 of 18

between the two sensor systems. Heart rate estimation using frequency domain analysis
maintains high efficiency and does not need labeled training data; however, these methods
usually estimate a heart rate over a period of a signal and hardly detect heartbeats. Thus,
developing an unsupervised beat-to-beat ECG heart rate estimation method that does not
rely on accurate locations of each heartbeat is more desirable and practical for daily heart
monitoring using wearable ECG sensors.
In this paper, the autoencoding technology is introduced into the feature extraction
of an ECG signal, and a beat-to-beat heart rate estimation based on Gaussian mixture
clustering is proposed for ECG signals from wearable sensors. Extending our previous
study [12], in order to further reduce the dependence on manual labeling, we introduce
the autoencoder Gaussian mixture model (AE-GMM) into an ECG heart rate estimation.
The deep one-dimensional convolutional autoencoder network is trained to extract ECG
signal heartbeat features by minimizing the reconstruction error; then, the adaptive number
of categories is introduced into Gaussian mixture clustering for heartbeat detection and
heart rate estimation. Compared with the existing ECG heart rate estimation methods, the
contributions of the proposed work are:
(1) As an unsupervised learning method, this method does not require precise manual
marking of the heartbeat locations, which can save labor costs, time, and is more practical.
Compared with the traditional signal processing methods, this method relies on the pow-
erful representation ability of the convolutional neural network to significantly improve
the accuracy of heart rate estimation. Its advantage lies in the fusion of three-dimensional
signals into one-dimensional features through the autoencoder, which can make full use of
the information of the three-lead ECG signals in an unsupervised manner to obtain more
prominent features.
(2) The collection of ECG signals is non-invasive through a wearable device, which
makes the collected signals inevitably vulnerable to interference and noise. Among them
are the distortion of the signal caused by motion and noise. The previous signal processing
methods are difficult to achieve satisfactory estimation performance on such noisy signals,
so we combined the deep autoencoder with Gaussian mixture modeling to propose this
unsupervised learning method to adaptively model the high-level heartbeat features and
achieve robust beat-to-beat heart rate estimation.
(3) The proposed method was fully verified on the recently developed variable
ECG sensors by the Suzhou Institute of Biomedical Engineering and Technology, Chi-
nese Academy of Sciences, providing a reliable beat-to-beat heart rate estimation solution
for wearable devices.

2. Related Work
ECG heartbeat waveforms maintain certain discriminative features that are instru-
mental for heart rate analysis, among which, the most important one is the detection of
QRS complex, namely, the R wave shown in Figure 2. However, in practice, the mixture
of system noise and/or body movements in ECG waveform make it quite challenging for
advanced analysis in the long-term. A number of effective methods have been proposed in
the literature. Table 1 lists some of the related methods, and we divide them into three cat-
egories. W. J. Tompkins proposed the Pan–Tompkins method, which is the most classic
R-wave detection method based on thresholding [13]. The Pan–Tompkins method consists
of two parts: the signal preprocessing stage and R-wave detection in the decision-making
stage, maintaining high real-time performance and high detection accuracy, while also re-
ducing the false detection of T-waves in the ECG recordings. Ramakrishnan et al. combined
adaptive moving window difference thresholding and front back difference thresholding
to detect the R wave [14]. Banerjee et al. applied discrete wavelet transform to learn
frequency features and then conducted classification of anteroseptal myocardial infarc-
tion [15]. Minami et al. designed a Fourier transform neural network for real-time detection
of ventricular tachyarrhythmia [16]. Sun et al. applied the morphological transform to learn
a discriminative template, then detected fiducial points in an ECG signal [17]. Kim et al.
Sensors 2021, 21, 7163 4 of 18

applied the difference operation method for R wave detection for a patch-type ECG remote
monitoring system [18]. Dewangan et al. combined wavelet and morphological coefficients
as input features and applied the neural network for arrhythmia classification [19]. Os-
owski et al. extracted the high-order statistics and Hermitian features of the QRS complex
and used a support vector machine for ECG heartbeat classification [20]. Other classic
ECG heart rate estimation methods include filter bank [21], phase space [22], the QRS
multilevel Teager energy operator (MTEO) [23], automatic multiscale-based peak detection
(AMPD) [24], and the UNSW QRS detection algorithm [25]. The filter bank [21] method
combines several band-pass filters in order to better delineate the QRS complexes. The
phase space [22] method adopts nonlinear phase space reconstruction recorded by ECG to
identify QRS complexes. MTEO [23] uses the multilevel Teager energy operator to locate
the QRS complex. UNSW QRS detection[25] applies QRS detection in the ECG signal
collected in the telehealth environment. The proposed method combined the filtered ECG
amplitude and derivative as input features and an adaptive thresholding was applied,
achieving more robustness for telehealth applications.

Figure 2. The standard waveform of ECG signal.

Traditional classification methods based on wavelet, morphological features, adaptive

thresholding, and others rely too much on feature engineering. The accuracy of classifi-
cation is limited by prior knowledge and often requires a lot of artificial features to assist
analysis, which is complex and time-consuming in practice. With the development of the
deep learning theory and the increasing power of parallel computing, end-to-end deep
learning based classification/regression methods emerge rapidly in ECG signal analysis.
Oh et al. introduced the one-dimensional convolutional neural network (CNN) and long
short-term memory (LSTM) for feature extraction and arrhythmia diagnosis from variable
length ECG signals [26]. Fotiadou et al. proposed to use convolutional neural network and
the LSTM network, which can fuse spatial and temporal information from multi-channel
fetal ECG signals to directly predict fetal heart rate [27]. Lee et al. proposed an algorithm
for detecting fetal QRS complexes in non-invasive fetal ECG (NI-FECG) signals based on
convolutional neural networks, which can reliably detect fetal QRS complexes without
separating the maternal ECG signals [28]. Because the NI-FECG signal contains a maternal
ECG signal with a larger amplitude than the fetal ECG signal, it can be seen from the side
that the deep network has a strong feature characterization ability.
Sensors 2021, 21, 7163 5 of 18

Deep learning methods often require adequate amounts of accurately labeled data
for training, which is usually very time- (and effort) consuming, and sometimes infeasible.
Therefore, it is more appealing to develop unsupervised feature learning algorithms for
ECG signal analysis. In [29], the K-singular value decomposition (K-SVD) method was
adopted for sparse large ECG data feature extraction and QRS classification. Ref. [30]
proposed a ECG heartbeat clustering method based on the unsupervised extreme learning
machine and decision rule. Ref. [31] proposed a heartbeat detection method based on
adaptive thresholding on the two-dimensional representation of ECG features obtained
from a number of independent detection methods.

Table 1. The related works of R wave detection.

Algorithm Category Year Method

1985 Pan–Tompkins [13]
1999 Fourier-transform neural network [16]
1999 Filter banks [21]
2002 Phase space [22]
2004 Support vector machine-based expert system [20]
2005 Morphological transform [17]
Traditional detection methods
2010 Discrete wavelet transform [15]
2011 Difference operation [18]
2015 Multilevel Teager energy operator (METO) [23]
2016 UNSW [25]
2016 Discrete wavelet transform and artificial neural network [19]
2017 Adaptive threshold [14]
2018 Combination of 1D-CNN and LSTM [26]
Supervised deep neural network methods 2018 Convolutional neural networks [28]
2020 Deep convolutional LSTM regression [27]
2012 Clustering and multimethod approach [31]
Unsupervised machine learning methods 2016 Advanced K-means clustering algorithm and K-SVD [29]
2016 Unsupervised ELM and decision rule [30]

3. Proposed Method
The goal of the proposed autoencoding Gaussian mixture model (AE-GMM) is to
estimate the locations of each heartbeat and compute the heart rate of a subject through
the ECG signal acquired by the wearable ECG device. Specifically, let X = {x1 , x2 , . . . , xn }
denote the input ECG data, X̂ denote the reconstructing ECG data and Y = {y1 , y2 , . . . , yn }
denote the hidden layer features extract from the encoder-decoder.

3.1. The 1D Convolutional Self-Encoder

Autoencoder is a feature extraction model composed of the neural network, which
consists of an encoder and decoder. The encoder is mainly responsible for the feature
representation of data, while the decoder is responsible for reconstructing the original data
through the characterization results; the parameters of the encoder–decoder are optimized
by minimizing the reconstruction error. In this paper, we constructed an incomplete one-
dimensional convolutional autoencoder by limiting the output dimension of the encoder to
be smaller than the original data. As shown in Figure 3, we used the incomplete learning
mechanism to constrain the autoencoder to capture the most significant features of the
ECG data.
Sensors 2021, 21, 7163 6 of 18

Figure 3. The structure of autoencoder.

For the input data x, the encoding process that maps it from the input layer to the
hidden layer can be represented as follows:

Y = f (Wen · X + ben ), (1)

where Wen and ben , respectively, are the connection weights and biases of the coding
module, Y is the encoded hidden layer features, and f (·) is the coding mapping function
of the input data.
The decoding process of reconstructing training samples X̂ from the hidden layer can
be expressed as follows:
X̂ = g(Wde · Y + bde ), (2)
where Wde and bde , respectively, are the connection weights and offsets of the decoding
module, and g(•) is the decoding mapping function of the encoding feature.
The training process can be expressed as minimizing the following loss function,

min Loss(X, X̂, θ) = (X − X̂)2 . (3)

θ

In Equation (3), Loss denotes the loss function, which is the mean square error in this
paper, and θ is the training parameters for the network.
(
0 x≤0
ReLU ( x ) = (4)
x x>0

The activation function used in this paper is Rectified Linear Unit (ReLU), and its
formula is shown in Equation (4). The activation function not only introduces nonlinear
factors into the network to improve the feature extraction ability of the model, it also
effectively avoids the gradient disappearance.

3.2. Gaussian Mixture Clustering

Gaussian mixture clustering is a probabilistic model that infers the mixture of a finite
number of Gaussian distributions generating the observed data. In theory, the Gaussian
mixture model can represent any complex distribution. In this paper, we use the Gaussian
mixture distribution to model the features extracted from the autoencoder for each ECG
heartbeat cycle.
For the multiple Gaussian distribution used to model the features Y, let p(Y | µ, Σ)
denote the probability density of the data, then the Gaussian mixture distribution composed
of K mixture components can be defined as,
K
pM = ∑ αk p(Y | µk , Σk ) (5)
k =1
Sensors 2021, 21, 7163 7 of 18

In Equation (5), p(Y | µk , Σk ) is Gaussian distribution density, and µk , Σk are the mean
vector and covariance matrix of the kth Gaussian mixture component, respectively. αk is
the mixing coefficient of the kth Gaussian mixture, which satisfies:
K
αk ∈ (0, 1], ∑ αk = 1 (6)
k =1

Assuming feature yi follows a certain Gaussian mixture, let Z be the label of the
Gaussian mixture component that generates the random variable yi . The prior probability
of Z is p( Z = k) = αk and the posterior distribution is shown as Equation (7).

p ( Z = k ) pM ( y i | Z = k )
pM ( Z = k | y i ) =
pM ( y i )
(7)
αk p(Y | µk , Σk )
=
∑kK=1 αk p(Y | µk , Σk )

Assuming that the Gaussian mixture distribution of Y is known, Gaussian mixture

clustering can divide the feature set into K clusters according to different Gaussian mixture
components. The clustering result of yi can be expressed as:

Cyi = arg max pM ( Z = k | yi ) (8)

k

For the features set Y = {y1 , y2 , . . . , yN }, the parameter learning of Gaussian mixture
distribution can be solved by expectation maximization (EM) algorithm [32]. Denoting the
number of Gaussian mixture components in the sample to be K, the optimization process
can be shown as follows:
(1) Initialize the model parameters of each Gaussian mixture component;
(2) Calculate the posterior probability γik = p M of each feature yi generated by each
mixed component according to Equation (7);
(3) Calculate the new mean vector µ of each Gaussian mixture component,

∑iN=1 γik yi
µ0k = , k = 1, 2, 3, . . . , K (9)
∑iN=1 γik
and the new covariance matrix

T
∑iN=1 γik yi − µ0k yi − µ0k


Σ0k = , k = 1, 2, 3, . . . , K (10)
∑iN=1 γik yi
and the new mixing coefficient

∑iN=1 γik
α0k = , k = 1, 2, 3, . . . , K (11)
N

(4) Update the parameters of each Gaussian mixture component;

(5) If the termination condition is satisfied, the final clustering classification is determined
according to Equation (8). If not, repeat Step (2)∼(4) until the termination condition
is satisfied.

3.3. Cluster Evaluation Function

In the proposed method, the Calinski–Harabasz index [33] was adopted as cluster
evaluation index. The Calinski–Harabasz clustering evaluation function is used to calculate
the scores of the clustering results under each category, and determine the heartbeat cluster
for heartbeat confirmation.
For clusters, the Calinski–Harabasz score is expressed as the ratio of the within-cluster
dispersion and the between-cluster dispersion. Higher Calinski–Harabasz scores are related
Sensors 2021, 21, 7163 8 of 18

to models with better defined clusters. The definition of Calinski–Harabasz index is shown
as follows:
tr( Bk )( N − K )
s(k) = , (12)
tr(Wk )(K − 1)
where N is the number of sample sets and K is the number of categories; tr (·) is the trace of
the matrix, Wk is the covariance matrix within the category, and Bk is the covariance matrix
between classes shown as follow:

Bk = ∑ nk (ck − c)(ck − c)T , (13)

k

where ck is the center of the kth cluster, c is the center of the whole feature set and nk is the
number of samples of the kth cluster.
It can be seen from the definition that the higher the Calinski–Harabasz index, the
better the clustering performance (dense and well separated). The heartbeat cluster is
confirmed by selecting the cluster with the highest Calinski–Harabasz score. As an unsu-
pervised performance evaluation index, it is very suitable for heartbeat cluster confirmation
for the proposed method in practice because of its low computational complexity and
relatively stable performance.

3.4. Algorithm Model

The proposed AE-GMM model includes three parts: heartbeat feature extraction,
heartbeat feature clustering and heartbeat cluster confirmation, and beat-to-beat heart rate
calculation, which are illustrated in the following parts and summarized in Algorithm 1.

Algorithm 1: AE-GMM
01 Find all peak locations of the three lead ECG signals.
02 Extract data segment centered at each peak with radius r = 45 as the input feature
for this candidate peak, resulting in feature tensor X in dimension [M, 91, 3],
where M is the total number of peaks, 2r + 1 = 91 is the total length of th input
heartbeat signals.
03 Standardize the input feature tensor X.
04 Train the autoencoding network and extract the compressed feature tensor Y with
dimension (M, 91, 1).
05 Perform Gaussian mixture clustering as described in Section 3.2.
06 Determine the heartbeat cluster according to the Calinski–Harabasz clustering scores.
07 Calculate the beat-to-beat heart rate based on the R wave locations.

(1) In the feature extraction section, the encoding and decoding parts of the convo-
lutional autoencoder are the one-dimensional convolutional and deconvolution network,
respectively. The structure of the model is shown in Figure 4, in which, the structure of the
convolutional submodule is shown in Figure 5. Through the operation of the convolution
kernel, as shown in Figure 5, the three-lead input signal is merged into one-dimensional,
with length 91 samples.
Sensors 2021, 21, 7163 9 of 18

Figure 4. The structure of AE-GMM, which contains encoder and decoder convolution layers and Gaussian mixture clustering.

Figure 5. The feature extraction structure of the encoder is composed of 6 layers of 1D CNN, batch
normalization (BN), and ReLU activation layer.

Q
1
L(x, x̂) =
Q ∑ (xi − xˆi )2 (14)
i =1

The loss function is shown in Equation (14), where xi and xˆi are the input and output
of the one-dimensional convolutional autoencoder, respectively, and Q is the batch size.
(2) In the Gaussian mixture clustering section, we first set the initial number of
categories in the range 3 ∼ 10, then traverse the range as the number of clusters for Gaussian
mixture clustering and obtain the corresponding clustering results. In the end, the Calinski–
Sensors 2021, 21, 7163 10 of 18

Harabasz score of each clustering result is computed in the clustering evaluation step; the
clustering result with the highest score is determined as the heartbeat cluster.
(3) The heartbeat locations are confirmed according to the heartbeat cluster and the
beat-to-beat heart rate is calculated. Specifically, mean time interval of the confirmed
heartbeats within a sliding window (with length Len, usually set to 30 s or 60 s) is firstly
computed and then transformed into beat/min, as shown in Figure 6. Suppose that there
are (m + 1) heartbeats in the sliding window and the time interval between every two
adjacent heartbeats in the sliding window is {t1 , t2 , · · · , tm }, the beat-to-beat heart rate(hr)
at time T is calculated as:
60
hr ( T ) = (15)
mean{t1 , t2 , . . . , tm }
Figure 7 shows the entire heart rate estimation process of the algorithm AE-GMM. The
purple dashed boxes in the figure represent the four key parts of the heart rate estimation
process. The first part is the sample construction. This process performs data normalization,
signal peak acquisition, signal interception, etc., as described in Algorithm 1. After the
autoencoder training is completed, the encoder in the autoencoder will be used to extract
the features of the samples, as shown in Figure 4. Then input the extracted features into
the Gaussian mixture clustering model to get the clustering result of the sample, as shown
in Figure 7, “0” in the classification result indicates that the sample is not a heartbeat, and
“1” indicates a true heartbeat. Finally, the heartbeat sequence diagram can be obtained
by combining the samples classified into the heartbeat and the position information of
the samples, and then the heart rate can be easily calculated, according to Figure 6 and
Equation (15).

Figure 6. Calculation of beat-to-beat heart rate. The blue curve is the ECG signal, and the yellow
solid dot is the reference position of the R wave.
Sensors 2021, 21, 7163 11 of 18

Figure 7. The complete heart rate estimation process of the AE-GMM method.

4. Experiment
The study was conducted according to the guidelines of the Declaration of Helsinki,
and approved by the Institutional Review Board, the Suzhou Institute of Biomedical
Engineering and Technology, Chinese Academy of Sciences.The proposed method was
evaluated on the ECG signal individually collected from healthy subjects by the wearable
ECG sensors. Informed consent was obtained from all subjects involved in the study. In
the training phase, the learning rate was 0.0005, the batch size was 128, and the Adam
optimizer was adopted for optimization. Since the proposed method was unsupervised
and no manually labeled information was required in the modeling process, it was not
necessary to divide the data into the training set and the test set. The training process stops
when the loss is less than a preset small threshold or the number of iterations is greater
than 1500 epochs. In the clustering section, the initial number of categories ranges from
3 to 10. The conditions for clustering to stop is that the number of iterations is greater than
100 or the parameter update amount is less than 0.001. The experiments were conducted
on a workstation with Intel i7 9700k CPU and NVIDIA 2080Ti GPU with 11 GB memory,
under Python3.6 and a TensorFlow 1.4 environment.

4.1. Data Preprocessing

Ten healthy subjects were selected to participate in the evaluation and the heart signals
were collected by the wearable ECG sensors shown in Figure 1. Each subject was required
to collect ECG signals for at least 3 min in a sitting state. The signal from each subject was
a three-channel lead data sampled at 100 Hz, and the data length is shown in Table 2.

Table 2. The signal length of subject data.

Subject Signal Length(s)

1 360
2 570
3, 5 420
4 540
6–10 180
Sensors 2021, 21, 7163 12 of 18

Due to the ECG signals containing noise, such as breathing, body movement, gastroin-
testinal movement, etc., a second-order Butterworth filter with a lower cut-off frequency of
0.4 Hz and upper cut-off frequency of 10 Hz was applied for noise removal.
In this experiment, we used the time domain segments after filtering from received
ECG signals as input features. Specifically, for each subject, because the peak position of
each lead signal cannot be a one-to-one correspondence, the peak locations from one lead
signal were labeled as the peak location of the entire ECG signal received by three channels
as candidate R peak locations (possible heartbeat locations). As shown in Figure 8, the
third row of signals was chosen as the reference signal. For each peak, we extracted a data
segment centered at each peak with radius r as the training feature for this peak location. In
practice, r was set to 45, resulting in 91 sample long features (corresponding to 0.91 s signal,
45 samples before and after the peak). This setting was verified in our previous work [34]
and was found to be the typical length of a heartbeat pattern. Figure 8 shows the three
lead ECG signals from subject no. 2, as an example, where filtered ECG signals collected
by three ECG transducers were plotted. The orange circles denote the signal peaks, and the
red X denote the example locations of R waves manually labeled by us and confirmed by a
medical expert. The peak of the R wave in the ECG signal is the most prominent feature.
Based on this, we assume that the heartbeat will only appear at the peak position.
In order to reduce the variance between input features, all input segments are stan-
dardized to have zero mean and unit variance.

Figure 8. Data segment diagram. The blue curve is the ECG three-lead signal, and the orange circle
represents the position of each signal segment intercepted by the point at the center with the left and
right radius r, and the red X are the ground truth of R wave locations.

4.2. Analysis of Experimental Results

In this section, we used a 30s sliding window to calculate beat-to-beat heart rates. The
mean absolute error between the calculated heart rate and the real heart rate was calculated.
The comparison methods, as Table 3 shows, include four unsupervised traditional QRS
wave monitoring methods for ECG signals provided by an open MATLAB toolbox [35],
Sensors 2021, 21, 7163 13 of 18

which are: PT [13], AMPD [24], MTEO [23], UNSW [25]; ensemble learning methods,
XGBoost [36], which have achieved good performance in many bioinformatics analysis;
Hilbert Transform (HT) [37]; and Evo-MIACE [12]. Specifically, HT and Evo-MIACE
are non-invasive heart rate estimation methods for the ballistocardiograms proposed by
our collaborators, which we believe could be good comparisons and provide insightful
evaluation on an ECG signal. For the HT method [37], the collected signal is decomposed
into the heartbeat component and the harmonic component in the frequency domain
by Hilbert transform; the heartbeat frequency component is used to compute heart rate.
Since HT requires a period of a signal for frequency analysis, the heart rate is estimated
through every 15-second long signal. Evo-MIACE [12] is a weakly supervised method that
combines evolutionary optimization and multiple instance learning to learn a heartbeat
“concept”. Then the heartbeat concept is used as a template to match the ECG signal and
detect heartbeat.
For comparison, Figure 9 shows the beat-to-beat heart rate estimated by the proposed
AE-GMM method and HT from subject no. 10. Figure 9a shows the beat-to-beat heart rate
estimated by the proposed AE-GMM method from subject no. 10, where one can clearly
see that the real time estimation of the heart rate, by the proposed methods, is very stable
and accurate; expect an overestimate at the beginning. Figure 9b displays the heart rate
estimated by the HT method, which is limited by the 15 s segment and not suitable for real
time beat-to-beat heart rate estimation.
In order to further examine and analyze the results, Figure 10 shows the detailed
heartbeats confirmed by the proposed method. From Figure 10, one can clearly see that
there exists only one false alarm heartbeat, where a non-R wave position is detected as an R
wave at about 21.8 s. Since the sliding window size is 30 s, the influence of the overestimate
of the heart rate lasts from 21.8 s to 51.8 s.

(a)

Figure 9. Cont.
Sensors 2021, 21, 7163 14 of 18

(b)

Figure 9. Comparison of heart rate estimation between AE-GMM and HT on subject no. 10. (a) Heart
rate estimation of AE-GMM on subject no. 10. (b) Heart rate estimation of HT on subject no. 10.

Figure 10. Heartbeat detection result of AE-GMM on subject no. 10. The figure shows the signals of
three leads from 20 to 25 s, and the horizontal axis of leads 2 and 3 are shifted downward by 3.5 and
7, respectively.

Table 4 summarizes the heart rate estimation error of AE-GMM and the comparison
algorithms evaluated individually on the ten subjects. The proposed AE-GMM method
achieved the overall best heart rate estimation performance, with average error smaller than
1 beat/min. It can be seen from Table 4 that some typical ECG heart rate estimation methods
Sensors 2021, 21, 7163 15 of 18

achieved excellent performance on several subjects, but failed (providing very large estimation
error) for wearable ECG lead signals with relatively poor signal quality. Some methods detect
Q, P, S, and T waves at the same time when detecting R waves, and use the positions of these
auxiliary waves to adjust the R waves detection. If the detection of these auxiliary waves is
inaccurate, the accuracy of the R wave confirmation will also be affected.
In order to gain deep insights on the learned heartbeat features, Figure 11 plots the
extracted wearable ECG heartbeat features for subject no. 10 by the proposed multi-layer
convolution autoencoding network, where one can clearly see that the features learned
by autoencoding maintain discriminative features and provide a guarantee for accurate
heartbeat separation.

Figure 11. Some features extracted from the autoencoding network.

In addition, we analyzed the effect of the sample length on the performance of heart
rate estimation. Table 5 shows the average heart rate estimation error under different
sample lengths of the ECG signals from five randomly selected subjects. We can see that,
as the sample length increases, the heart rate estimation error tends to decrease first and
then increase. Because when the sample length is very short, each sample cannot contain a
complete heartbeat, which makes the algorithm unable to effectively extract the features
of the heartbeat. The algorithm has the best performance when the sample length is set
to 91 sampling points (0.91 s). Since the sampling frequency applied to the ECG signal is
100 Hz, setting the length of each sample to 91 sampling points (0.91 s) can make almost
every sample contain a full heartbeat cycle, which will be more conducive to the extraction
of the ECG signal R wave features.
We also investigated the effects of different depths of autoencoder layers on the
performance of heart rate estimation. Table 6 shows the average heart rate estimation
error when the number of convolutional layers of the autoencoder was set to 3, 6, 9 on
the ECG data of five randomly picked subjects. It can be seen from Table 6 that when
the number of convolutional layers was 3, there was an under-fitting phenomenon in the
autoencoder training, which led to a larger heart rate estimation error. When the number of
convolutional layers was 9, the heart rate estimation performance significantly improved
compared to the 3-layer setting, but it was almost the same as the number of convolutional
layers as 6. This shows that when the number of convolutional layers is small, increasing
the number of convolutional layers can effectively improve the estimation performance.
However, when the number of convolutional layers increases to a certain level, as the
number of convolutional layers increases, the improvement in estimation performance
Sensors 2021, 21, 7163 16 of 18

will become insignificant. And it needs to be noted that the increase in the number of
convolutional layers will lead to increased computational complexity of model training,
reduce the time efficiency of model training, and increase the risk of the model overfitting.

Table 3. The comparison algorithms of this paper.

Method Abbreviation Method

PT Pan–Tompkins [13]
AMPD Automatic peak detection [24]
METO Multilevel Teager energy operator [23]
UNSW UNSW [25]
HT Hilbert transform [37]
Evo-MIACE Evolutionary optimized multiple instance concept learning [12]
XGBoost eXtreme gradient boosting [36]

Table 4. Performance of AG-GMM and comparisons across the 10 subjects, bold for the best, underline for the second best, standard
deviations smaller than 0.01 are denoted as 0.00.

Mean Absolute Error (beat/min)

Subject
PT [13] AMPD [24] METO [23] UNSW [25] HT [37] Evo-MIACE [12] XGBoost [36] AE-GMM
1 2.27 0.09 16.16 22.32 3.45 2.28 ± 0.15 2.24 0.05
2 1.00 1.55 0.92 1.18 3.53 2.19 ± 0.59 2.93 1.07
3 0.15 0.17 0.04 24.65 1.65 0.92 ± 0.50 1.30 0.00
4 0.09 0.16 0.00 0.05 2.67 1.25 ± 1.34 2.18 0.19
5 0.58 0.58 0.62 0.42 1.54 3.29 ± 0.70 2.47 0.96
6 0.45 20.02 0.80 2.56 1.36 1.59 ± 0.10 2.51 1.58
7 2.91 20.18 2.56 6.50 0.73 2.31 ± 0.29 1.63 1.48
8 0.00 0.71 0.76 0.00 3.71 3.76 ± 1.22 1.95 0.13
9 0.00 1.77 0.00 0.56 2.07 6.27 ± 2.64 3.49 0.91
10 12.64 0.28 0.30 0.12 2.91 0.00 ± 0.00 1.64 0.36
Total average 2.01 4.55 2.22 5.84 2.36 2.39 2.23 0.67

Table 5. Heart rate estimation error under different sample lengths. The bold are the sample length
set in our method and the heart rate estimation error.

Sample length 31 61 121 151 91

Average error 2.49 2.02 1.31 1.29 0.51

Table 6. Heart rate estimation error under different number of convolutional layers of autoencoder.
The bold are the number of layers set in our method and the heart rate estimation error.

Number of layers 3 9 6
Average error 1.01 0.54 0.50

5. Conclusions
This paper introduces the adaptive mechanism of the number of clusters in the
clustering, and automatically selects the optimal number of clusters by setting the clustering
evaluation function, which also ensures the high accuracy of the final heartbeat detection.
In this paper, the deep autoencoding strategy is introduced into the feature extraction of
ECG wearable sensor signals, and a heartbeat detection and beat-to-beat heart rate estimation
method is proposed based on convolutional autoencoding and Gaussian mixture clustering.
By training convolutional autoencoding network, the high-level R wave features are extracted
in an unsupervised manner, which could be applied for Gaussian mixture clustering for
Sensors 2021, 21, 7163 17 of 18

heartbeat confirmation. Experimental results show that the proposed method has achieved
superior performance with an average mean absolute error as 0.67 beat/min, providing a
robust beat-to-beat heart rate estimation solution for wearable ECG systems in practice.

Author Contributions: All of the authors made significant contributions to this work. J.Z. and C.J.
designed the ECG system and collected the data. J.C. and D.H. devised the approach and wrote
the paper. Y.L. and S.G. conceived and designed the experiments; H.Z. and W.Z. conducted the
experiments and analyzed the data. All authors have read and agreed to the published version of
the manuscript.
Funding: This work was supported in part by the Suzhou Science and Technology Project (Re-
search on Remote Real-time Multi-Vital Sign Wearable Monitoring Smart Clothes, no. SYG201906),
the Science and Technology Cooperation High-Tech Industrialization Special Project funded by
Jilin Province, the Chinese Academy of Sciences (Intelligent Evaluation System for Mental Stress
State based on the Perception of Multiple Physiological Information, no. 2020SYHZ0043) and the
Natural Science Foundation of Shaanxi Province under Grant (No. 2019ZDLGY03-02-02).
Institutional Review Board Statement: The study was conducted according to the guidelines of the
Declaration of Helsinki, and approved by the Institutional Review Board the Suzhou Institute of
Biomedical Engineering and Technology, Chinese Academy of Sciences.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Virani, S.S.; Alonso, A.; Benjamin, E.J.; Bittencourt, M.S.; Callaway, C.W.; Carson, A.P.; Chamberlain, A.M.; Chang, A.R.; Cheng, S.;
Delling, F.N.; et al. Heart Disease and Stroke Statistics—2020 Update: A Report From the American Heart Association. Circulation
2020, 141, e139–e596. [CrossRef]
2. Junaid, M.J.A.; Kumar, R. Data Science And Its Application In Heart Disease Prediction. In Proceedings of the 2020 International
Conference on Intelligent Engineering and Management, London, UK, 17–19 June 2020; pp. 396–400.
3. Ghaleb, F.A.; Kamat, M.; Salleh, M.; Rohani, M.F.; Hadji, S.E. Motion Artifact Reduction Algorithm Using Sequential Adaptive
Noise Filters and Estimation Methods for Mobile ECG. In Proceedings of the International Conference of Reliable Information
and Communication Technology, Johor Bahru, Malaysia, 23–24 April 2017; pp. 116–123.
4. Yoo, J.; Yan, L.; Lee, S.; Kim, H.; Yoo, H.J. A Wearable ECG Acquisition System with Compact Planar-Fashionable Circuit
Board-Based Shirt. IEEE Trans. Inf. Technol. Biomed. 2009, 13, 897–902. [CrossRef] [PubMed]
5. Pham, T.; Lau, Z.J.; Chen, S.H.A.; Makowski, D. Heart Rate Variability in Psychology: A Review of HRV Indices and an Analysis
Tutorial. Sensors 2021, 21, 3998. [CrossRef]
6. Dittmar, A.; Axisa, F.; Delhomme, G.; Gehin, C. New Concepts and Technologies in Home Care and Ambulatory Monitoring.
Stud. Health Technol. Inform. 2004, 108, 9–35.
7. Sung, M.; Marci, C.; Pentland, A. Wearable feedback systems for rehabilitation. J. Neuroeng. Rehabil. 2005, 2, 17. [CrossRef]
8. Heilman, K.J.; Porges, S.W. Accuracy of the LifeShirt® (Vivometrics) in the detection of cardiac rhythms. Biol. Psychol. 2007,
75, 300–305. [CrossRef] [PubMed]
9. Curone, D.; Secco, E.L.; Tognetti, A.; Loriga, G.; Dudnik, G.; Risatti, M.; Whyte, R.; Bonfiglio, A.; Magenes, G. Smart Garments for
Emergency Operators: The ProeTEX Project. IEEE Trans. Inf. Technol. Biomed. 2010, 14, 694–701. [CrossRef] [PubMed]
10. Lee, Y.D.; Chung, W.Y. Wireless Sensor Network based Wearable Smart Shirt for Ubiquitous Health and Activity Monitoring.
Sens. Actuators B Chem. 2009, 140, 390–395. [CrossRef]
11. Chen, C.C.; Lin, S.Y.; Chang, W.Y. Novel Stable Capacitive Electrocardiogram Measurement System. Sensors 2021, 21, 3668.
[CrossRef] [PubMed]
12. Cheng, J.; Zhong, J.; Wang, H.; Tang, X.; Jiao, C.; Zhou, H. Evolutionary Optimized Multiple Instance Concept Learning for
Beat-to-Beat Heart Rate Estimation from Electrocardiograms. In Proceedings of the 2020 13th International Congress on Image
and Signal Processing, BioMedical Engineering and Informatics, Chengdu, China, 17–19 October 2020; pp. 727–732.
13. Pan, J.; Tompkins, W.J. A Real-time QRS Detection Algorithm. IEEE Trans. Biomed. Eng. 1985, BME-32, 230–236. [CrossRef]
14. Ramakrishnan, S.; Yogeswaran, R. Design and Analysis of Feature Extraction Algorithm for ECG Signals using Adaptive Thresh-
old Method. In Proceedings of the 2017 Trends in Industrial Measurement and Automation, Chennai, India, 6–8 January 2017;
pp. 1–8.
15. Banerjee, S.; Mitra, M. ECG Feature Extraction and Classification of Anteroseptal Myocardial Infarction and Normal Subjects
using Discrete Wavelet Transform. In Proceedings of the 2010 International Conference on Systems in Medicine and Biology,
Kharagpur, India, 16–18 December 2010; pp. 55–60.
16. Minami, K.I.; Nakajima, H.; Toyoshima, T. Real-time Discrimination of Ventricular Tachyarrhythmia with Fourier-Transform
Neural Network. IEEE Trans. Biomed. Eng. 1999, 46, 179–185. [CrossRef]
Sensors 2021, 21, 7163 18 of 18

17. Sun, Y.; Chan, K.L.; Krishnan, S.M. Characteristic Wave Detection in ECG Signal using Morphological Transform. BMC Cardiovasc.
Disord. 2005, 5, 1–7. [CrossRef]
18. Kim, M.S.; Cho, Y.C.; Seo, S.T.; Son, C.S.; Kim, Y.N. Auto-Detection of R Wave in Electrocardiography for Patch-Type ECG Remote
Monitoring System. Biomed. Eng. Lett. 2011, 1, 180. [CrossRef]
19. Dewangan, N.K.; Shukla, S. ECG Arrhythmia Classification using Discrete Wavelet Transform and Artificial Neural Network.
In Proceedings of the 2016 IEEE International Conference on Recent Trends in Electronics, Information & Communication
Technology, Bangalore, India, 20–21 May 2016; pp. 1892–1896.
20. Osowski, S.; Hoai, L.T.; Markiewicz, T. Support Vector Machine-based Expert System for Reliable Heartbeat Recognition. IEEE
Trans. Biomed. Eng. 2004, 51, 582–589. [CrossRef]
21. Afonso, V.X.; Tompkins, W.J.; Nguyen, T.Q.; Luo, S. ECG Beat Detection using Filter Banks. IEEE Trans. Biomed. Eng. 1999,
46, 192–202. [CrossRef]
22. Lee, J.W.; Kim, K.S.; Lee, B.; Lee, B.; Lee, M.H. A Real Time QRS Detection Using Delay-Coordinate Mapping for the Microcon-
troller Implementation. Ann. Biomed. Eng. 2002, 30, 1140–1151. [CrossRef]
23. Sedghamiz, H.; Santonocito, D. Unsupervised Detection and Classification of Motor Unit Action Potentials in Intramuscular Elec-
tromyography Signals. In Proceedings of the 2015 E-Health and Bioengineering Conference, Iasi, Romania, 19–21 November 2015;
pp. 1–6.
24. Scholkmann, F.; Boss, J.; Wolf, M. An Efficient Algorithm for Automatic Peak Detection in Noisy Periodic and Quasi-Periodic
Signals. Algorithms 2012, 5, 588–603. [CrossRef]
25. Khamis, H.; Weiss, R.; Xie, Y.; Chang, C.W.; Lovell, N.H.; Redmond, S.J. QRS Detection Algorithm for Telehealth Electrocardiogram
Recordings. IEEE Trans. Biomed. Eng. 2016, 63, 1377–1388. [CrossRef]
26. Oh, S.L.; Ng, E.Y.; Tan, R.S.; Acharya, U.R. Automated Diagnosis of Arrhythmia using Combination of CNN and LSTM
Techniques with Variable Length Heart Beats. Comput. Biol. Med. 2018, 102, 278–287. [CrossRef] [PubMed]
27. Fotiadou, E.; Xu, M.; van Erp, B.; van Sloun, R.J.; Vullings, R. Deep Convolutional Long Short-Term Memory Network for Fetal
Heart Rate Extraction. In Proceedings of the 2020 42nd Annual International Conference of the IEEE Engineering in Medicine &
Biology Society (EMBC), Montreal, QC, Canada, 20–24 July 2020; pp. 1–4.
28. Lee, J.S.; Seo, M.; Kim, S.W.; Choi, M. Fetal QRS detection based on convolutional neural networks in noninvasive fetal
electrocardiogram. In Proceedings of the 2018 4th International Conference on Frontiers of Signal Processing (ICFSP), Poitiers,
France, 24–27 September 2018; pp. 75–78.
29. Balouchestani, M.; Krishnan, S. Advanced K-means clustering algorithm for large ECG data sets based on a collaboration of
compressed sensing theory and K-SVD approach. Signal Image Video Process. 2016, 10, 113–120. [CrossRef]
30. Annam, J.R.; Surampudi, B.R. AAMI Based ECG Heart-Beat Time-Series Clustering Using Unsupervised ELM and Decision
Rule. In Proceedings of the 2016 International Conference on Information Technology, Bhubaneswar, India, 22–24 December 2016;
pp. 137–141.
31. Zazula, D. Optimization of Heartbeat Detection based on Clustering and Multimethod Approach. In Proceedings of the 2012
Annual International Conference of the IEEE Engineering in Medicine and Biology Society, San Diego, CA, USA, 28 August–
1 September 2012; pp. 5–8.
32. Dempster, A.P.; Laird, N.M.; Rubin, D.B. Maximum Likelihood from Incomplete Data via the EM Algorithm. J. R. Stat. Soc. Ser. B
(Methodol.) 1977, 39, 1–22.
33. Caliński, T.; Harabasz, J. A Dendrite Method for Cluster Analysis. Commun.-Stat.-Theory Methods 1974, 3, 1–27. [CrossRef]
34. Jiao, C.; Su, B.Y.; Lyons, P.; others. Multiple instance dictionary learning for beat-to-beat heart rate monitoring from ballistocar-
diograms. IEEE Trans. Biomed. Eng. 2018, 65, 2634–2648. [CrossRef] [PubMed]
35. Sedghamiz, H. BioSigKit: A Matlab Toolbox and Interface for Analysis of Biosignals. J. Open Source Softw. 2018, 3, 671. [CrossRef]
36. Chen, T.; Guestrin, C. Xgboost: A Scalable Tree Boosting System. In Proceedings of the 22nd ACM Sigkdd International
Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA, 13–17 August 2016; pp. 785–794.
37. Rosales, L.; Su, B.Y.; Skubic, M.; Ho, K. Heart Rate Monitoring using Hydraulic Bed Sensor Ballistocardiogram. J. Ambient. Intell.
Smart Environ. 2017, 9, 193–207. [CrossRef]