See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/282420582 Assessment of Biofeedback Training for Emotion Management Through Wearable Textile Physiological Monitoring System Article in IEEE Sensors Journal · December 2015 DOI: 10.1109/JSEN.2015.2470638 CITATIONS READS 40 431 5 authors, including: S.C. Mukhopadhyay Wanqing Wu Macquarie University Shenzhen Institutes of Advanced Technologies (SIAT) 480 PUBLICATIONS 6,345 CITATIONS 85 PUBLICATIONS 730 CITATIONS SEE PROFILE Sandeep Pirbhulal Chinese Academy of Sciences 54 PUBLICATIONS 461 CITATIONS SEE PROFILE Some of the authors of this publication are also working on these related projects: Brain Korean 21 Century View project Tumors View project All content following this page was uploaded by Sandeep Pirbhulal on 15 July 2017. The user has requested enhancement of the downloaded file. SEE PROFILE IEEE SENSORS JOURNAL, VOL. 15, NO. 12, DECEMBER 2015 7087 Assessment of Biofeedback Training for Emotion Management Through Wearable Textile Physiological Monitoring System Wanqing Wu+ , Heye Zhang+ , Sandeep Pirbhulal, Subhas Chandra Mukhopadhyay, Fellow, IEEE, and Yuan-Ting Zhang Abstract— Negative emotion has a wide range of pernicious impacts on people, ranging from the failure in real-time task performance to the development of chronic health conditions. An unobtrusive wearable biofeedback system for personalized emotional management has been designed and presented in this paper. The system integrated heart rate variability (HRV) biofeedback to wearable biosensor platform, which could function both as an early stress warning system as well as a visual interface to manipulate subject’s affective state. The designed and developed system would help subject to transform the negative emotion state into positive through real-time HRV biofeedback training. The results indicated that the real-time HRV biofeedback is significantly effective in cases of negative emotion. With the aid of the developed biofeedback system, the subhealth subjects could transform heart rhythm from negative emotion to positive emotion-related oscillation mode. Index Terms— HRV biofeedback, wearable biosensor, heart rate variability, heart rhythm pattern, emotional management. I. I NTRODUCTION IOFEEDBACK has been described as a ‘psychophysiological mirror’, allowing the patients to monitor and learn from physiological signals produced by the body [1], [2]. The clinical efficacy of biofeedback has been investigated in a range of psychiatric disorders, including anxiety [3], depression [4], and schizophrenia [5]. Biofeedback was B Manuscript received June 15, 2015; revised August 9, 2015; accepted August 10, 2015. Date of publication September 2, 2015; date of current version October 8, 2015. This work was supported in part by in part by the Key Laboratory for Health Informatics, Chinese Academy of Sciences, and in part by the Science Technology and Innovation Committee of Shenzhen for Research Projects (CXZZ20140909004122087 and JCYJ20140901003939025). The associate editor coordinating the review of this paper and approving it for publication was Prof. Ignacio R. Matias. (Wanqing Wu and Heye Zhang contributed to this work equally.) (Corresponding author: Subhas Chandra Mukhopadhyay.) W. Wu, H. Zhang, and S. Pirbhulal are with the Institute of Biomedical and Health Engineering, and also with the Key Laboratory for Health Informatics in Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences, Shenzhen 518055, China (e-mail: wq.wu@siat.ac.cn; hy.zhang@siat.ac.cn; sandeep@siat.ac.cn). S. C. Mukhopadhyay is with the School of Engineering and Advanced Technology, Massey University, Palmerston North 4442, New Zealand (e-mail: s.c.mukhopadhyay@massey.ac.nz). Y. T. Zhang is with the Key Laboratory for Health Informatics, Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences, Shenzhen 518055, China, and also with the Joint Research Centre for Biomedical Engineering, Department of Electronic Engineering, The Chinese University of Hong Kong, Hong Kong (e-mail: ytzhang@ee.cuhk.edu.hk). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/JSEN.2015.2470638 found to be cost-effective on all dimensions reviewed with cost/benefit ratios ranging between 1:2 and 1:5, which provided high-efficiency in clinical settings, such as the reduction in medical care costs to patients, decrease in frequency and duration of hospital stays and re-hospitalization, more important, biofeedback treatment was verified for decreasing in mortality and increasing in quality of life [6], [7]. Despite of the 27% of people in Europe suffering from mental health problems each year (Lancet Global Mental Health Group, 2007), 74% of these people receive no pharmaceutical or traditional psychological treatment from mental health-care services, often due to multiple barriers in accessing such services [1]. For example, previous biofeedback treatment can be very expensive because it might require a large amount of training and practice under the professional supervision. Such treatment processing can hardly conduct by the people in home environment. Although physiological measurement devices have been widely used in clinical biofeedback settings for many years, little work has focused on automated stress interventions under the non-clinical environment. Hence, different types of biofeedback strategies, such as heart rhythm coherence feedback, oscillatory feedback and resonance frequency training feedback (RFT) [8]–[13] have recently been used for self-regulation of central and autonomic nervous system. And, more and more researches have been dedicated to raise living quality in terms of health through designing and fabricating intelligent sensors that are characterized by unobtrusive and non-invasive. Advances in the fields of electronics and instrumentation technology, together with novel textile materials and novel textile-electronic integration techniques have boosted the development and implementation of garments based on sensors for wearable bio-signal measurement systems [1]. As a result of extensive and numerous research efforts, several wearable and textile-enable monitoring systems for the performance of non-invasive measurements have been designed and implemented. For example, a Zigbee smart noninvasive wearable physiological parameters monitoring device use biofeedback is developed to monitor physiological parameters, such as temperature and heart rate. The system consists of an electronic device that is worn on the wrist and finger, by an at-risk person [14]–[16]. To assess the elderly activities at home in real-time, a low cost, robust, flexible and efficiently home monitor system is developed using Wireless Sensor Network (WSN) [15], [17]–[19]. The simplicity of binary 1530-437X © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. 7088 IEEE SENSORS JOURNAL, VOL. 15, NO. 12, DECEMBER 2015 Fig. 1. Architecture of proposed biofeedback system. passive infrared (PIR) sensors allowed us to collect the movements of the elderly, which are supportive in the daily living (ADL) assessment and the subsequent wellness determination process [20]–[23] Also experiments showed visual relation between emotions (i.e. happy, sad, angry and neutral) and physiological signals, which are obtained from a heart rate sensor, a skin temperature sensor and a skin conductance sensor [24], [25]. With the increasing availability of sensors and methods for detecting stress, some unique features of unobtrusive and wearable devices due to recent advances in sensing, networking and data fusion have broadened the range of biofeedback application. It is has become one trend to integrate intellectual sensors and biofeedback theory to exercise stress management. Consequently, the studies on the integration of wearable technology and biofeedback procedures for psychiatric disorders is one undoubtedly demand to overcome the limitation of conventional biofeedback treatment. Hence, the purpose of this study was to implement a new biofeedback methodology for personalized stress management, which is based on theoretical basis of biofeedback strategies and the construction of an unobtrusive wearable biosensor platform, the block diagram of system architecture is displayed in Fig.1. Given that the primary application of the system is to help the user to self-correct and regulate his/her physiological functions both in static and dynamic conditions, the requirements taken into account for the system design are: i) wearability and unobtrusiveness; ii) low-power consumption and low-cost; iii) flexibility and easy integration with other sensors and actuators; iv) easy maintenance and update of components; and v) reliability. Through the convenient, reliable and accurate wearable platform, we can instruct the person to take the proper action to release the stressful mental functions using collected multi-scenario physiological data and biofeedback theory. II. S YSTEM D ESIGN The model of our biofeedback methodology is built on the theoretical basis of the biofeedback closed loop model, autonomic cardiovascular control model. Furthermore, respiration training closed loop is also incorporated to theoretical model. The structure of proposed theoretical model in our biofeedback methodology is illustrated in the Fig.2. This proposed three-layer system model is directly associated with three interventions involved in biofeedback Fig. 2. The theoretical three-layer closed loop model of proposed system. Fig. 3. The implemented principle framework of proposed theoretical model. closed loop model: human, machine and environment (individual/external environment), and accordingly brought out three challenges in this thesis: Correlations between physiological markers and stress level; Functionality and performance of the biofeedback system; Effectiveness of respiration training strategy. The brain and the spinal cord comprise the central nervous system (CNS), which through the Autonomic Nervous System (ANS) and peripheral innervation of organs and glands controls the heart’s electrical activity, gland secretion, blood pressure, and respiration function among others to preserve the homeostasis of the organism. Analysis of ANS activity regarding Sympathetic Nervous System (SNS) and Parasympathetic Nervous System (PNS) is a common practice for assessing stress [26]. In order to realizing stress evaluation and feedback training simultaneously, we designed an unobtrusive wearable biofeedback system for recording physiological variables (ECG and Respiration) that would allow studying the response of the ANS during stressful tasks in a non-invasive manner, the principle framework of this study is illustrated in Fig.3. A. Description of System Architecture The system architecture follows the principle framework of novel biofeedback methodology as shown in Fig.3, which basically contains a network of sensor nodes, cross-platform intelligent software system, personalized emotion assessment and biofeedback training model. The analog front end (AFE) of proposed system is based on the capacitive sensing principle, and comprises of textile based polymer fabrics electrode array, general structure analog WU et al.: ASSESSMENT OF BIOFEEDBACK TRAINING FOR EMOTION MANAGEMENT 7089 Where C [F] is capacitance of the coupling, R[] is resistance of the inserted insulator (clothes) and f [Hz] is frequency of the body source signal. Since R is so high in this study that it can be regarded as infinity, and C can be represented by using coupling area S [m2 ], distance d [m] between textile electrode and the skin, and permittivity ε [F/m] of the insulator (clothes), so the equation (1) can be expressed by equation (2): d 1 × (2) Z R=∞ = 2π f εS Fig. 4. Block diagram of hardware platform based on capacitive sensing principle. conditioning module, universal digital signal processing unit and wireless transmitter module, which amplifies, filters and then relays the signals to the computer. Emotion assessment algorithm and biofeedback mechanism are installed in the PC/smart phone, which initiates and loops the audio/video cues till the abnormal stress level is detected. Such system is capable of driving different actuators as biofeedback generators and provides a promising methodology for emotion management, and has the potential to integrate more sensing nodes as different actuators for biofeedback regularization. B. Hardware Platform Description Because our system needs to record ECG and respiratory data continuously for at least one week, we chose a conductive textile material as the electrodes in our system, which can extract breathing activity and ECG signal from individual simultaneously. The AFE consisted of a differential separation filter and a common signal conditioning part, and the block diagram of whole hardware platform was illustrated in Fig. 4. 1) Principle and Construction of Capacitive Sensing for ECG and Respiration Measurement: The proposed approach of capacitive sensing by using wearable textile electrode for ECG and breathing activity measurement is an expansion of the principle of the capacitive (or insulator) electrode, the coupling is composed of a conductive fabric electrode, clothes and the skin of the subject, as shown in Fig.4. According to the equivalent circuit elements, impedance Z [] of coupling is expressed by equation (1): Z= R 1 + (2π f C R)2 = 1 1 R2 + (2π f C)2 (1) As we known, the capacitive sensing is susceptible to body motion, which is considered as the geometric parameters S and d have been changed, so the obtained signals had contained a periodic variation involving low frequency component and had seemed to be caused by breathing activity in the position of chest. Based on all these facts, a separation filter (1Hz Differential Separation Filter) is employed in our proposed measuring system to divide the original signal into a high frequency component containing ECG signal (>1Hz) and a low frequency component including respiratory signal (<1Hz), which was constructed of two sets of subtracter, amplifiers and integrators according to DC suppression circuit. The signal conditioning unit with general structure for sensing ECG signal consisted of an instrumentation amplifier, a high-pass filter (HPF), a low pass filter (LPF) and two inverting amplifiers. The circuit elements of the HPF and the LPF were designed in order to obtain a cutoff frequency of 0.5 and 40Hz, respectively. Although electrocardiograph for diagnostic purpose requires a bandwidth from 0.01 to 100Hz, we narrowed the bandwidth of the developed part for decreasing noise interference. The signal conditioning unit for sensing breathing activity consisted of an instrumentation amplifier, a high pass filter and an inverting amplifier. 2) Construction of Capacitive Driven-Right-Leg (DRL) for ECG Measurement: For the reduction of common-mode noise level in capacitive ECG measurement, a driven-rightleg (DRL) circuit was employed in this study. Generally, ECG measurements are carried out by the differential method, which is very effective in reducing the common mode noise. However, in real measurement, this method does not work as well as expected due to the large asymmetry in the two signal path. Therefore, DRL circuit is a must-have component in AFE for common mode noise reduction, which is usually attached on the right leg and connected to the instrument ground. And the common electrode is driven by the inverse of common mode noise component, and then the impedance between the body and the sensing instrument is reduced. In this study, we have carefully considered the material and wearable modality of textile electrodes, and designed two rectangular lead electrodes with 3cm width, and a rectangular third electrode for DRL circuit with 5cm width. The DRL was applied to our system and the equivalent circuit of the system has been illustrated in Fig.5 (a) and (b). From the equivalent circuit in fig.5 (b), VO = −G × VC M , G = 2R F RE (3) 7090 IEEE SENSORS JOURNAL, VOL. 15, NO. 12, DECEMBER 2015 Fig. 5. (a) The configuration of textile electrode-based wearable system including driven right leg circuit and (b) is the equivalent circuit of (a). Fig. 6. Equivalent circuit for skin-electrode impedance [27]. VO = VC M − (R O + Z C G ) × i 1 (4) where VC M is the common mode voltage, G is the resistor ratio of the inverting amplifier (A3), ZCG is the impedance of textile electrode, and R O is a current limiting resistor. From equation (3) and (4), R O + Z CG (5) G+1 Equation (5) shows that if the third textile electrode is connected to a driven right leg circuit instead, and if G > Ro/ZCG , then this reduces the effective resistance to common RCM and thus reduces common mode voltage VC M . 3) Technical and Electronic Method for Variation of Skin-Electrode Impedance: In this study, how varying skin-electrode interface impedance affects the bio-signal sensing is also an important issue with regard to textile electrode-based ECG measurement. In the literature [27], Swanson and Webster have developed a model for skin-electrode impedance, as shown in Fig. 6. The model can be expressed by equation (6): w2 Cd2 Rd2 Rd |Z (w)| = Rs + −j 1 + w2 Cd2 Rd2 1 + w2 Cd2 Rd2 VC M = RC M × i 1, RC M = S (6) d Where Rs represents the electrolyte gel (if any), sweat, and the underlying skin tissue, generally, the Rs is constant for an individual. Rd stands for the resistance that occurs between the skin and electrode during charge transfer. In this equation, while the capacitance Cd represents the electrical charge between the electrode and skin, which can be calculated by using coupling area S, distance d between textile electrode and the skin, and permittivity ε. In this work, we have used a smart textile electrode with good comfort and characteristic, and can w = 2π f, Cd = ε × adapt to different shapes of body with single size. Therefore, the permittivity ε, and the resistance of fabric electrode (Rd ) is also constant, furthermore, the textile electrodes’ size (S) have enlarged and been configured to the rectangular with 3cm width for achieving reliable capacitive coupling [27]. Consequently, in the equation (6), the skin-electrode impedance varies with the distance between skin and conductive textile electrodes and the external pressure applied to the electrode surface. Therefore, in this work, the textile electrode we opted for ECG measurement has elastic characteristics to make a better contact between the electrode and the skin that leads to smaller skin-electrode impedance. And, the electronic method we selected to decrease the effect of varying skin electrode impedance as time passes is to install a high input resistance pre-amplifier (active electrode A1, A2) on the rear side of the conductive textile electrode. The pre-amplifier converts the high impedance between the fabric electrode and skin to low output impedance required by subsequent circuitry. The voltage variations on the skin sensed by the two fabric electrode through the capacitive couplings are amplified by each pre-amplifier. C. Software Platform Description The software platform has three main components: a physiological signal-processing unit, a physiological signalanalysis unit, and a biofeedback regulation unit. The signals collected in this work can be quite weak and easily destroyed by a variety of noises, which can come from Electromagnetic interference of power line, poor quality of contact between the electrode and the skin, baseline wander caused by respiration, noises generated by hardware platform or electrosurgical instrument, movement of body position, and any other possible noise sources. Most of these noises cannot be filtered out completely over the hardware-processing unit. Therefore, it is necessary to filter out these noises as much as we can in the software platform. We adopted Butterworth Notch Filter (BNF) and finite impulse response (FIR) band-pass filter to eliminate the power line interference and baseline wander, and a novel multi-scale mathematical morphology (3M) filter to reduce the impact of the nonlinear noises caused by poor electrode contact and motion artifacts [28], [29]. In the end, we use a differential operation method (DOM) [30] to smooth and normalize the physiological signal collected by our hardware platform in this signal-processing unit. In the physiological signal-analysis unit, we use an adaptive QRS waveform identification algorithm developed in our previous work [31] to extract RR interval from ECG signal. The results generated by this algorithm can be used to measure the HRV and heart rhythm pattern. We analyze the heart rhythm pattern using the morphological inspection and quantitative method [32] simultaneously. The calculated time domain (Mean RR, SDNN, rMSSD, pNN50), frequency domains (VLF, absolute and normalized LF, HF, total power), geometric (TINN, HRV TI) and nonlinear measures (Poincare plot & Detrended Fluctuation Analysis) of HRV are obtained according to the standards of measurement, WU et al.: ASSESSMENT OF BIOFEEDBACK TRAINING FOR EMOTION MANAGEMENT 7091 proposed by the Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology [33], [34], which describes the detail of physiological correlates of HRV and calculation methods. The biofeedback module was built on the theoretical basis of RFT and heart rhythm pattern. We utilized these two efficient biofeedback mechanisms simultaneously to alter the physiological and emotional state through respiratory control. The biofeedback module basically included two parts: an algorithm module and a GUI module (integrated into the bio-signal monitoring module). There are two procedures in the algorithm designed for biofeedback application in this work. In the first procedure, cyclic measurements were collected at different frequencies of respiratory (4–7 breaths per minute and in increments of 0.5), and the respiratory rate was recorded when HRV and CR reached the maximum gains, and heart rate curve reached maximum oscillation amplitude. After collecting cyclic measurements, the subject in this experiment was asked to gradually adjust breathing to his/her resonant frequency with the help from our biofeedback regulation unit. Finally, the subjects were asked to keep the depth of respiration in an approximately constant frequency. In the second procedure, the subject in this experiment might use the resonant frequency in self-regulation to control emotion and stress. III. E XPERIMENT R ESULTS AND D ISCUSSION In this study, textile electrodes and hardware platform were integrated into an elastic belt that could be comfortable worn around thorax or abdomen (chest strap or waist belt) for physiological monitoring and biofeedback training, furthermore, we also afforded a reflective photo-sensor to detect PPG signal for the future study, which appears as the form of wrist strap. Based on such wearable components, we could construct a miniaturized body sensor network (BSN) [35] in combination with the base station node (or BT dongle) and smart terminals to provide an inexpensive, unobtrusive, and unsupervised ambulatory monitoring during normal daily activities. To aim at existed challenging issues of BSN, this proposed wearable biofeedback system was constructed by aforementioned smart textile electrodes, biosensor node platform, and associated biosensors, meanwhile, a battery board has been well-designed to provide selectable power solutions for the purpose of low consumption. In addition, a base station node with Bluetooth function was developed to aggregates information from the distributed biosensor nodes and ultimately conveys it across existing networks to other stakeholders or intelligent terminals, such as smart phone, PAD and notebook. Inherently, this proposed wearable biofeedback system was based on sophisticated circuit design, configuration and integration; it not only facilitates non-expert users for self-emotion management, but also allows the caregiver to select appropriate biofeedback strategies for individuals with enough user-specific options. Consequently, the proposed wearable biofeedback is of great importance to be a crucial component in BSN framework for realizing ubiquitous, affordable and customized self emotionmanagement. The system implementation was detail illustrated in Fig.7. Fig. 7. System implementation of proposed biofeedback methodology. A. Evaluation of HW/SW System We performed the in-situ dynamic respiration experiments in a closed room with constant temperature controlled by the air conditioner. All the subjects were required to lie down at supine positions when they wore our device on the wrist or elbow. During the experiment, we also used one commercially available ECG measurement kit named “PolyG-A” (LAXTHA Inc.) to collect the signal under the same conditions, and its results would be compared to our device so as to evaluate the performance of our device. In conducted experiment, the grounds of Poly G-A system and designed biofeedback system were connected, so to measure the respiration signal and ECG signal l from electrodes (Lead II) under the 512 Hz sampling rate. Based on the waveforms of ECG and respiration signals acquired by both systems shown in Fig.8, we can concluded that the quality of ECG and respiration signals was very close to that from PolyG-A. We also displayed the HRV curves, respiration curves, and PSD analysis results, which were generated by Telescan software packaged (LAXTHA Inc.) and our physiological signal analyzer respectively, the comparative results as shown in Fig.8 demonstrated the performance of wearable biofeedback system is very close to the commercial bed-side medical monitoring system. Because the low-power consumption has been one of significant challenges in wireless sensor network, we collect the information of the power consumption by measuring the 7092 IEEE SENSORS JOURNAL, VOL. 15, NO. 12, DECEMBER 2015 TABLE I C OMPARISONS OF R ESPIRATORY PATTERN D URING T HREE C ONDITIONS Fig. 8. The performances comparison of LAXTHA PolyG-A and our biofeedback system. aforementioned three battery power supply solutions for our experiment respectively: one normal battery (9V, 350mAh, 3.15Wh), a portable solar battery (3.7V, 350mAh, 1.295Wh), and a button cell battery (3V, 250mAh, 0.75Wh). During the experiment, our hardware platform ran the ECG and respiration sensors in a full capacity. Therefore, the data throughput of two channels was 64 kbps with sampling rate and length in 512Hz and 12bit respectively. Consequently, our platform could run about 10 hours with the support of 9V battery, 5 hours with the support of solar battery and 3.5 hours with the support of button cell battery respectively for ECG and respiration data collection. B. Evaluation of Biofeedback Mechanism 1) Experiment Procedure: 15 healthy, right-handed subjects (all male ranging from 20 to 28 years of age, with no history of cardiovascular disease) recruited from a local university took part in this pilot study, which has been approved by the Institutional Review Board (IRB) of Pusan National University Hospital. After the informed written consent was collected from all the participants, they were asked to complete questionnaires for examining their psycho-physiological assessment. All subjects were tested under standard conditions between 1:00 and 5:00 P.M., at a room temperature of 22-26 °C after abstaining from smoking and coffee consumption for 6 hours before participation in the experiment, and asked to wear the device comfortably with regular electrodes. Lead II ECG signals and respiration signal were collected for each subject simultaneously. Visual stimuli pictures from the International Affective Picture System (IAPS) [36] were selected randomly to evoke negative emotional responses. The experiment procedure includes three sessions: 1) Measurement of HRV parameters, oscillate amplitude of heart rate, and coherence ratio under the rest state (baseline condition, BL); 2) In the absence of resonant frequency-based respiration biofeedback, evaluating the physiological responses (HRV, Respiration and heart rhythm pattern) under the negative emotion (NE_nBF); 3) Evaluating the physiological responses under the negative emotion (NE_BF) with resonant frequencybased respiration biofeedback training. In this pilot study, HRV temporal parameters included: SDNN, which is a global index of HRV and reflects all long-term components and circadian rhythms responsible for variability in the recording period; rMSSD, which reflects parasympathetic nerve activity, and the HRV triangular index (HRV TI), which serves as an estimate of the overall HRV. In the HRV PSD analysis, the low frequency (LF, 0.04-0.15Hz) and high frequency (HF, 0.15-0.4Hz) components were measured in normalized units; the former reflects respiration and is associated solely with activity in the parasympathetic nervous system, whereas the latter is associated with sympathetic and parasympathetic modulation of RR intervals and is strongly influenced by baroreflex activity. Correspondingly, the LF/HF ratio was calculated to estimate the sympatho-vagal balance. Moreover, we also measure the coherence ratio (CR) that reflects the degree of ANS synchronization. Pair wise comparisons were made to compare the parameters difference between the NE_BF and NE_nBF by using the paired t-test. And one-way ANOVA for repeated measures was also performed in order to evaluate the effects of the resonant frequency-based respiration biofeedback on HRV parameters and heart rhythm pattern. Probability values less than 0.05 were considered statistically significant (SPSS v19.0, Chicago, IL, USA). C. Analyses of Respiratory Pattern Table 1 demonstrates the mean value of respiratory rate, expiration (Exp.) time, inspiration time (Insp.) and the ratio of expiration to inspiration time (E/I ratio). As expected, the three trials differed significantly in the respiration measures, illustrating that the experimental manipulation of the respiration pattern was successful. In the NE_nBF trail, the respiratory rate was 15.3cpm, whereas in the NE_BF trail the respiratory rate was 6.5cpm. Compared with BL, the respiratory rate was significantly increased during NE_nBF WU et al.: ASSESSMENT OF BIOFEEDBACK TRAINING FOR EMOTION MANAGEMENT 7093 TABLE II C OMPARISONS OF H EART R ATE VARIABILITY U NDER T HREE C ONDITIONS Fig. 9. Three representative samples of heart rhythm pattern analysis and corresponding coherence ratio (CR). (p<0.05) mainly due to the influences of negative emotion, and obviously decreased during NE_BF (p<0.01) condition attributed to conscious respiratory control. The expiration times were approximately 3-fold and 2-fold larger in the NE_BF than NE_nBF and BL condition. Although there was no obvious difference between BL and NE_nBF trials, the inspiration time during NE_BF was significantly higher than NE_nBF (p<0.05). Moreover, the repeated measures ANOVA showed that volitional respiratory control could change the respiratory pattern greatly during NE_BF trial. D. Analyses of Heart Rhythm Pattern The pattern of heart rhythm looks like periodic sine waves during process of converting the negative emotion into positive emotion. Fig.9 showed three typical examples of heart rhythm patterns derived from the subject. The morphological differences of these heart rhythm patterns happened during NE_BF, NE_nBF and BL trials. The negative emotion would introduce a disjointed and unpredictable heart rhythm pattern together with low oscillation-amplitude during NE_nBF. However, more regular and coherent heart rate rhythm, with a concomitant increase in oscillation amplitude, has been observed in the NE_BF trial, which reflected increased ANS synchronization and manifested a sine-wave-like oscillating mode of heart rhythm (positive emotion). E. Analyses of Heart Rate Variability In Table 2, compared with BL condition, significantly increased CR (p<0.01), SDNN (p<0.01), rMSSD (p<0.01), HRV TI (p<0.05) and normalized HF (p<0.05) have been observed during NE_BF trial, and decreased significantly in the NE_nBF trial correspondingly. Also, with or without the resonant frequency-based respiratory control, these measures differed greatly in the presence of negative stimuli. Significantly increases in CR, SDNN and HRV TI during NE_BF trial implied enhanced total HRV and responsiveness of the cardiac autonomic system to psycho-physiological stimuli, and increased rMSSD and normalized HF, together with significantly decreased LF (p<0.05) indicated the inhibition of sympathetic nervous system coupled with the activation of parasympathetic nervous system. In the NE_nBF trial, except normalized LF, all of the time domain, frequency domain and geometric measures were much lower than BL section, which indicated diminished total HRV and over-activation of sympathetic output flow. With resonant frequency based respiratory control, the subjects could speed up recovery of the cardiovascular system from the adverse impacts of negative emotions by themselves. Moreover, the LF/HF ratio during NE_BF trial was much lower than NE_nBF also demonstrated that resonant frequency based respiratory control produced a higher sympatho-vagal balance under the negative emotion. The repeated measures ANOVA further revealed that resonant respiratory frequency was capable of profoundly affecting the time domain, frequency domain and geometric measures in the presence of negative emotion, and shifting heart rhythm toward a positive emotion related oscillation mode. IV. C ONCLUSIONS The wearable device with biofeedback function has been one emerging researching application, which can integrate modern biofeedback theories into the state-of-the-art sensing and BSN technologies. The wearable device developed in this work is generic with low power-cost and low complexity, and it can be expanded to build more wearable biofeedback-training devices. This pilot study used wearable training patterns and resultant HRV to exercise the biofeedback functions. HRV dynamics can be easily affected by the human’s physiological and emotional states. For example, positive and negative emotions can be distinguished by smooth or erratic heart rhythm patterns, respectively. Particularly, heart rhythms become more irregular during the experience of negative emotions, which can imply less synchronization in the reciprocal action between the sympathetic and parasympathetic branches of the autonomic nervous system. Correspondingly, 7094 IEEE SENSORS JOURNAL, VOL. 15, NO. 12, DECEMBER 2015 sustained positive emotion was associated with greater synchronization between the two branches, which can indicate a highly coherent heart rhythm pattern. We used the realtime morphological characteristic of HRV curve as a visual tool to evaluate the effects of resonance frequency based respiratory feedback training. In addition, the CR was adopted to assess the destructive influence of negative emotion on heart rhythm pattern, which derived from spectral measures of HRV. Also, CR is very effective in quantifying the performance of resonance frequency respiratory training. The results of statistical analysis in our experiments can confirm that the relationship between negative emotion and HRV. These findings were also supported by the lower value of CR, SDNN, rMSSD, HRV-TI, and higher LF/HF ratio under NE_nBF trial compared with BL and NE_BF trail (as shown in Table 2). Meanwhile, the graphical analysis results provided the evidence for the effectiveness of resonance frequency based respiratory training. As shown in Fig.9, during the NE_BF trial, the heart rhythm pattern of individuals were more regular and coherent (higher CR value) compared to NE_nBF trial, which implied that resonance respiratory training could decrease the harmful influences of negative emotion to ANS modulation. Therefore, all analysis results suggested that resonant frequency based respiratory combined with heart rhythm pattern feedback was appropriate to decrease sympathetic arousal, increase parasympathetic activity, and enhance overall capability of ANS modulation in the presence of negative emotion, including time domain, frequency domain, quantified heart rhythm pattern and morphological recognition of HRV curves. However, the number of samples used in this study was small. We will collect a large-scale experimental data using wearable textile physiological monitoring system in the future study. BSN is the network of several associated sensor nodes on, inside or around human body to monitor vital signals, which has showed enormous potential for ubiquitous and low-cost healthcare. Therefore, we might investigate more advanced infrastructure of the BSN-based biofeedback systems by using multiple physiological parameters, and explore the data confidentiality and security (biometric security) based on HRV features to attain privacy during physiological information processing in BSN. Competing Interests The authors declare that they have no conflict of interest. R EFERENCES [1] P. L. Schoenberg and A. S. David, “Biofeedback for psychiatric disorders: A systematic review,” Appl. Psychophysiol. Biofeedback, vol. 39, no. 2, pp. 109–135, 2014. [2] S. Elmer and L. Jäncke, “Intracerebral functional connectivity-guided neurofeedback as a putative rehabilitative intervention for ameliorating auditory-related dysfunctions,” Frontiers Psychol., vol. 5, Oct. 2014, Art. ID 1227. [3] E. Peper, R. Harvey, and N. Takabayashi, “Biofeedback an evidence based approach in clinical practice,” Jpn. J. Biofeedback Res., vol. 36, no. 1, pp. 3–10, 2009. [4] A. J. Beckham, T. B. Greene, and S. Meltzer-Brody, “A pilot study of heart rate variability biofeedback therapy in the treatment of perinatal depression on a specialized perinatal psychiatry inpatient unit,” Arch. Women’s Mental Health, vol. 16, no. 1, pp. 59–65, 2013. [5] J. E. Walker and R. Lawson, “FP02 beta training for drug-resistant depression—A new protocol that usually reduces depression and keeps it reduced,” J. Neurotherapy, vol. 17, no. 3, pp. 198–200, 2013. [6] M. S. Schwartz and F. Andrasik, Biofeedback: A Practitioner’s Guide, 3rd ed. New York, NY, USA: Guilford Press, 2003. [7] L. F. Nicolas-Alonso and J. Gomez-Gil, “Brain computer interfaces: A review,” Sensors, vol. 12, no. 2, pp. 1211–1279, 2012. [8] G. E. Prinsloo, H. G. Rauch, and W. E. Derman, “A brief review and clinical application of heart rate variability biofeedback in sports, exercise, and rehabilitation medicine,” Phys. Sportsmed., vol. 42, no. 2, pp. 88–99, 2014. [9] P. Lehrer, “History of heart rate variability biofeedback research: A personal and scientific voyage,” Biofeedback, vol. 41, no. 3, pp. 88–97, 2013. [10] P. Lehrer et al., “Protocol for heart rate variability biofeedback training,” Biofeedback, vol. 41, no. 3, pp. 98–109, 2013. [11] E. G. Vaschillo, B. Vaschillo, R. J. Pandina, and M. E. Bates, “Resonances in the cardiovascular system caused by rhythmical muscle tension,” Psychophysiology, vol. 48, no. 7, pp. 927–936, 2011. [12] W. Wu, Y. Gil, and J. Lee, “Combination of wearable multi-biosensor platform and resonance frequency training for stress management of the unemployed population,” Sensors, vol. 12, no. 10, pp. 13225–13248, 2012. [13] A. P. Sutarto, M. N. A. Wahab, and N. M. Zin, “Heart rate variability (HRV) biofeedback: A new training approach for operator’s performance enhancement,” J. Ind. Eng. Manage., vol. 3, no. 1, pp. 176–198, 2010. [14] K. Malhi, S. C. Mukhopadhyay, J. Schnepper, M. Haefke, and H. Ewald, “A Zigbee-based wearable physiological parameters monitoring system,” IEEE Sensors J., vol. 12, no. 3, pp. 423–430, Mar. 2012. [15] N. Samanta, A. Chanda, and C. RoyChaudhuri, “An energy efficient, minimally intrusive multi-sensor intelligent system for health monitoring of elderly people,” Int. J. Smart Sens. Intell. Syst., vol. 7, no. 2, pp. 762–780, 2014. [16] X. Zhenghua et al., “The implementation for the intelligent home control system based on the android and Zigbee,” Int. J. Smart Sens. Intell. Syst., vol. 7, no. 3, pp. 1095–1113, 2014. [17] N. K. Suryadevara and S. C. Mukhopadhyay, “Wireless sensor network based home monitoring system for wellness determination of elderly,” IEEE Sensors J., vol. 12, no. 6, pp. 1965–1972, Jun. 2012. [18] R. C. King, L. Atallah, B. Lo, and G.-Z. Yang, “Development of a wireless sensor glove for surgical skills assessment,” IEEE Trans. Inf. Technol. Biomed., vol. 13, no. 5, pp. 673–679, Sep. 2009. [19] N. K. Suryadevara, S. C. Mukhopadhyay, R. Wang, and R. K. Rayudu, “Forecasting the behavior of an elderly using wireless sensors data in a smart home,” Eng. Appl. Artif. Intell., vol. 26, no. 10, pp. 2641–2652, 2013. [20] N. K. Suryadevara and S. C. Mukhopadhyay, “Determining wellness through an ambient assisted living environment,” IEEE Intell. Syst., vol. 29, no. 3, pp. 30–37, May/Jun. 2014. [21] N. K. Suryadevara, M. T. Quazi, and S. C. Mukhopadhyay, “Intelligent sensing systems for measuring wellness indices of the daily activities for the elderly,” in Proc. 8th Int. Conf. Intell. Environ. (IE), Jun. 2012, pp. 347–350. [22] S. C. Mukhopadhyay, “Wearable sensors for human activity monitoring: A review,” IEEE Sensors J., vol. 15, no. 3, pp. 1312–1330, Mar. 2015. [23] H. Jing, “Coverage holes recovery algorithm based on nodes balance distance of underwater wireless sensor network,” Int. J. Smart Sens. Intell. Syst., vol. 7, no. 4, pp. 1890–1907, 2014. [24] M. T. Quazi, S. C. Mukhopadhyay, N. K. Suryadevara, and Y. M. Huang, “Towards the smart sensors based human emotion recognition,” in Proc. IEEE Int. Instrum. Meas. Technol. Conf. (I2MTC), May 2012, pp. 2365–2370. [25] S. C. Mukhopadhyay, N. K. Suryadevara, and R. K. Rayudu, “Are technologies assisted homes safer for the elderly?” in Pervasive and Mobile Sensing and Computing for Healthcare. Berlin, Germany: Springer-Verlag, 2013, pp. 51–68. [26] E. Vaschillo, P. Lehrer, N. Rishe, and M. Konstantinov, “Heart rate variability biofeedback as a method for assessing baroreflex function: A preliminary study of resonance in the cardiovascular system,” Appl. Psychophysiol. Biofeedback, vol. 27, no. 1, pp. 1–27, 2002. [27] H. Miller and D. C. Harrison, Biomedical Electrode Technology: Theory and Practice. New York, NY, USA: Academic, 1974. [28] S. Prakash and V. Venkatesh, “Real time monitoring of ECG signal using PIC and Web server,” Int. J. Eng. Technol., vol. 5, no. 2, pp. 1047–1053, 2013. WU et al.: ASSESSMENT OF BIOFEEDBACK TRAINING FOR EMOTION MANAGEMENT [29] C. Zhu and F. Tian, “An ECG detection algorithm using wavelet and autocorrelation transform,” in Proc. Int. Conf. Wireless Commun. Signal Process. (WCSP), 2013, pp. 1–6. [30] Y.-C. Yeh and W.-J. Wang, “QRS complexes detection for ECG signal: The difference operation method,” Comput. Methods Programs Biomed., vol. 91, no. 3, pp. 245–254, 2008. [31] W. Wu and J. Lee, “Improvement of HRV methodology for positive/negative emotion assessment,” in Proc. 5th Int. Conf. Collaborative Comput., Netw., Appl. Worksharing (CollaborateCom), Nov. 2009, pp. 1–6. [32] S. Kim et al., “Heart rate variability biofeedback, executive functioning and chronic brain injury,” Brain Injury, vol. 27, no. 2, pp. 209–222, 2013. [33] F. X. Gamelin, S. Berthoin, and L. Bosquet, “Validity of the polar S810 heart rate monitor to measure R-R intervals at rest,” Med. Sci. Sports Exercise, vol. 38, no. 5, pp. 887–893, 2006. [34] M. Malik, “Heart rate variability standards of measurement, physiological interpretation, and clinical use,” Circulation, vol. 93, no. 3, pp. 354– 381, 1996. [35] S. Pirbhulal, “An efficient biometric-based algorithm using heart rate variability for securing body sensor networks,” Sensors, vol. 15, no. 7, pp. 15067–15089, 2015. [36] N. Suvorov, “Psychophysiological training of operators in adaptive biofeedback cardiorhythm control,” Spanish J. Psychol., vol. 9, no. 2, pp. 193–200, 2006. Wanqing Wu received the B.S. degree in computer science and technology from Hunan Normal University, Hunan, China, in 2004, the M.E. degree in computer science from Chongqing University, Chongqing, China, in 2007, and the Ph.D. degree from the Pusan National University of Computer Enigneering, Korea, in 2013. He is currently an Associate Professor with the Institute of Biomedical and Health Engineering, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China. He is also with the Key Laboratory for Health Informatics of the Chinese Academy of Sciences, China. His research interests include wearable medical device, biomedical signal sensing and processing, body sensor networks, physiological health informatics, and biofeedback theory. Heye Zhang received the B.S. and M.E. degrees from Tsinghua University, Beijing, China, in 2001 and 2003, respectively, and the Ph.D. degree from the Hong Kong University of Science and Technology, Hong Kong, in 2007. He is currently an Associate Professor with the Institute of Biomedical and Health Engineering, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China. He is also with the Key Laboratory for Health Informatics of the Chinese Academy of Sciences, China. His research interests include cardiac electrophysiology and cardiac image analysis. Sandeep Pirbhulal received the B.E. degree in telecommunication engineering from the Mehran University of Engineering and Technology, Pakistan, in 2011, and the M.S. degree in telecommunication and networks from the Karachi Institute of Economics and Technology, Pakistan, in 2014. He is currently pursuing the Ph.D. degree with the Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences. He has authored four SCI journal papers and one book chapter. His current research includes biomedical signal processing and biometrics-based information security. View publication stats 7095 Subhas Chandra Mukhopadhyay (M’97–SM’02– F’11) received the (Hons.) degree from the Department of Electrical Engineering, Jadavpur University, Kolkata, India, the master’s degree in electrical engineering from the Indian Institute of Science, Bangalore, India, the Ph.D. (Eng.) degree from Jadavpur University, and the Dr.Ing. degree from Kanazawa University, Kanazawa, Japan. He is currently a Professor of Sensing Technology with the School of Engineering and Advanced Technology, Massey University, Palmerston North, New Zealand. He has over 25 years of teaching and research experiences. He has authored or co-authored four books and over 300 papers in different international journals, conferences, and book chapters. He has edited 12 conference proceedings. He has also edited 12 special issues of international journals as a lead Guest Editor and 20 books, in which 18 are with Springer-Verlag. He has delivered 246 seminars as keynote, invited, tutorial, and special lectures in 24 countries. His fields of interest include sensors and sensing technology, instrumentation, wireless sensor networks, electromagnetics, control, electrical machines, and numerical field calculation. He received numerous awards throughout his career and attracted over NZ U.S. $4.1 million on different research projects. He is a Fellow of the Institution of Engineering and Technology, U.K., and the Institution of Electronics and Telecommunication Engineers, India. He is a Topical Editor of the IEEE S ENSORS J OURNAL, and an Associate Editor of the IEEE T RANSACTIONS ON I NSTRUMENTATION AND M EASUREMENTS . He is the Co-Editor-in-Chief of the International Journal on Smart Sensing and Intelligent Systems. He was the Technical Program Chair of ICARA 2004, ICARA 2006, and ICARA 2009. He was the General Chair/Co-Chair of ICST 2005, ICST 2007, the IEEE ROSE 2007, the IEEE EPSA 2008, ICST 2008, the IEEE Sensors 2008, ICST 2010, the IEEE Sensors 2010, ICST 2011, ICST 2012, ICST 2013, and ICST 2014. He has organized the IEEE Sensors Conference 2009, Christchurch, New Zealand, in 2009, as the General Chair. He will organize the Ninth ICST, Auckland, New Zealand, in 2015. He was a Distinguished Lecturer of the IEEE Sensors Council from 2010 to 2013. He is the Founding and Ex-Chair of the IEEE Instrumentation and Measurement Society New Zealand Chapter. He is the Chair of the IEEE IMS Technical Committee 18 on Environmental Measurements. Yuan-Ting Zhang received the Ph.D. degree from the University of New Brunswick, Canada, in 1990. He was a Research Associate and an Adjunct Assistant Professor with the University of Calgary from 1989 to 1994. He joined The Chinese University of Hong Kong (CUHK) as a Lecturer in 1994, became an Associate Professor in 1996, and a Professor in 2002. He serves as the Director of the Joint Research Centre for Biomedical Engineering. At CUHK, he has developed and teaches courses including biomedical modeling, medical instruments and sensors, and telemedicine techniques and applications. His research activities have focused on the development of biomodels and biosignal processing techniques to improve the performance of medical devices and biosensors, in particular, for telemedicine. His work has been published in several books, over 20 scholarly journals, and numerous international conference proceedings. Dr. Zhang held various positions in professional organizations. He served as the Technical Program Chair of the 20th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBS). He was the Chairman of the Biomedical Division with the Hong Kong Institution of Engineers in 1996/97 and 2000/01. He was elected and served as an AdCom Member of the IEEE-EMBS in 1999, and the Vice President of the IEEE EMBS in 2000 and 2001. He serves as an Associate Editor of the IEEE T RANSACTIONS ON B IOMEDICAL E NGINEERING and the IEEE T RANSACTIONS ON M OBILE C OMPUTING, and an Editorial Board Member for the Book Series of Biomedical Engineering (Wiley and IEEE Press). He is the General Conference Chair of the 27th Annual International Conference of IEEE EMBS.