PPG Features Research Articles

Photoplethysmographic Prediction of the Ankle-Brachial Pressure Index through a Machine Learning Approach

Cardiovascular disease is a leading cause of death. Several markers have been proposed to predict cardiovascular morbidity. The ankle-brachial index (ABI) marker is defined as the ratio between the ankle and the arm systolic blood pressures, and it is generally assessed through sphygmomanometers. An alternative tool for cardiovascular status assessment is Photoplethysmography (PPG). PPG is a non-invasive optical technique that measures volumetric blood changes induced by pulse pressure propagation within arteries. However, PPG does not provide absolute pressure estimation, making assessment of cardiovascular status less direct. The capability of a multivariate data-driven approach to predict ABI from peculiar PPG features was investigated here. ABI was measured using a commercial instrument (Enverdis Vascular Explorer, VE-ABI), and it was then used for a General Linear Model estimation of ABI from multi-site PPG in a supervised learning framework (PPG-ABI). A Receiver Operating Characteristic (ROC) analysis allowed to investigate the capability of PPG-ABI to discriminate cardiovascular impairment as defined by VE-ABI. Findings suggested that ABI can be estimated form PPG (r = 0.79) and can identify pathological cardiovascular status (AUC = 0.85). The advantages of PPG are simplicity, speed and operator-independency, allowing extensive screening of cardiovascular status and associated cardiovascular risks.

Cuffless Single-Site Photoplethysmography for Blood Pressure Monitoring.

One in three adults worldwide has hypertension, which is associated with significant morbidity and mortality. Consequently, there is a global demand for continuous and non-invasive blood pressure (BP) measurements that are convenient, easy to use, and more accurate than the currently available methods for detecting hypertension. This could easily be achieved through the integration of single-site photoplethysmography (PPG) readings into wearable devices, although improved reliability and an understanding of BP estimation accuracy are essential. This review paper focuses on understanding the features of PPG associated with BP and examines the development of this technology over the 2010–2019 period in terms of validation, sample size, diversity of subjects, and datasets used. Challenges and opportunities to move single-site PPG forward are also discussed.

Characterization of Artifact Signals in Neck Photoplethysmography.

The neck is a very attractive measurement location for multimodal physiological monitoring, since it offers the possibility of extracting clinically relevant parameters, which cannot be obtained from other body locations, such as lung volumes. It is for this reason that obtaining PPG from the neck would be of interest. PPG signals, however, are very susceptible to artifacts which greatly compromise their quality. But the extent of this is going to depend on, the nature of the artifacts and the strength of the sensed signal, both of which are location dependent. This paper presents for the first time the characterization of artifacts affecting neck PPG signals. Neck PPG data was recorded from 19 participants, who performed ten different activities to deliberately introduce common artifacts. 41 PPG features were extracted and statistically analyzed to investigate which ones showed the greatest ability to differentiate normal PPG from each artifact. A customized minimum Redundancy Maximum Relevance (mRMR) feature selection approach was implemented, to select the top 10 features. Artifacts caused by Swallowing, Yawning and Coughing exhibited larger Spectral Entropy, Average Power and smaller Spectral Kurtosis, than normal PPG. Head movement artifacts, also demonstrated highly disordered and noisy frequency spectra, and were characterized by having larger and irregular time domain features. In addition, the analysis showed that different respiratory states that could be of clinical interests, such as presence of apneas, were also distinguishable from sources of interference. These findings are important for the development of PPG denoising algorithms and subsequent obtention of biomarkers of interest, or alternatively for applications where the events of interest are the artifacts themselves.

Robust blood pressure estimation from finger photoplethysmography using age-dependent linear models

Objective: This work aims to develop an efficient and robust age-dependent multiple linear regression (MLR) model to estimate blood pressure (BP) from a single-source photoplethysmography (PPG) and biometrics, which could be embedded in the microcontroller of pulse oximeters. Approach: Hemodynamic features were extracted from the PPG signal using its waveform, derivatives, and biometrics. Whole-based, feature-based, and fusion models were evaluated and compared for different age groups. Their performance was tested using 1086 subjects with a leave-one-subject-out cross-validation. The improvement by adding biometrics and the long-term calibration effect were investigated in detail. The relative importance of each feature was compared between different age groups and the implication was discussed. Main results: The fusion model achieved the best performance in subjects with well-defined PPG features, whereas the feature-based method was better suited for subjects with damped signals. Adding age significantly improved both systolic BP (SBP) and diastolic BP (DBP) estimation accuracy for older subjects (> 50 years old) with well-defined features, while it only improved diastolic BP accuracy for older subjects with damped signals. For younger subjects (≤ 50 years old), the contribution of age was very small. A simple subtraction of subject-specific calibration factors significantly reduced biometric-related errors, which also improved the linearity of BP estimation. The relative importance analysis of input features suggests that separate models are indeed necessary for different age groups with different signal qualities, especially for DBP estimation in older subjects. Significance: This study shows a reasonable BP estimation accuracy with age-dependent MLR models, which may help to equip current pulse oximeters with additional functionalities.

Effect of Laryngeal Mask Airway Insertion on Parameters Derived From Catacrotic Phase of Photoplethysmography Under Different Concentrations of Remifentanil.

Background: Some parameters have been extracted from photoplethysmography (PPG) with a good relativity with nociception, but without encouraging results in qualifying the balance of nociception-anti-nociception (NAN). The features of PPG have not been thoroughly depicted and more prospective univariate parameters deserve to be explored. The aim of this study was to investigate the ability of parameters derived from catacrotic phase of PPG to grade the level of analgesia. Methods: 45 patients with ASA I or II were randomized to receive a remifentanil effect-compartment target controlled infusion (Ceremi) of 0, 1, or 3 ng/ml, and a propofol effect-compartment target controlled infusion to maintain an acceptable level of hypnosis with state entropy (SE) at 40~60. Laryngeal mask airway (LMA) insertion was applied as a noxious stimulus. Five diastole-related parameters, namely diastolic interval (DI), diastolic slope (DS), the minimum slope during catacrotic phase (DSmin), the interval between DSmin and its nearest trough (DTI), and area difference ratio (ADR), were extracted. Pulse beat interval (PBI) was calculated as a reference parameter. Results: LMA insertion elicited a significant variation in all parameters except ADR during Ceremi of 0 and 1 ng/ml. Compared to PBI (prediction probability (n}{}text{P}_{text {K}}n) = 0.796), the parameters of DI, DS, and DTI presented a better consistence with the level of anti-nociceptive medication, with n}{}text{P}_{text {K}}n of 0.825, 0.822, and 0.822 respectively. Conclusion: The features extracted from catacrotic phase of PPG, including DI, DS, and DTI, could provide a promising potential to qualify the balance of NAN.

The potential of photoplethysmogram and galvanic skin response in emotion recognition using nonlinear features.

Recently, developing an accurate automatic emotion recognition system using a minimum number of bio-signals has become a challenging issue in "affective computing." This study aimed to propose a reliable system by examining nonlinear dynamics of photoplethysmogram (PPG) and galvanic skin response (GSR). To address this goal, two strategies were adopted. First, the efficiency of each signal in valence/arousal based emotion categorization was examined. Then, the proficiency of a hybrid feature, by combining both GSR and PPG features was studied. Lyapunov exponents, lagged Poincare's measures, and approximate entropy were extracted to characterize the irregularity and chaotic behavior of the phase space. To discriminate two levels of arousal and two levels of the valence, a probabilistic neural network (PNN) with different sigma adjustment parameter was examined. The results showed that the phase space geometry and consequently, the signal dynamics are influenced by the emotional music video. Additionally, distinctive patterns of the phase space behavior were observed under the influence of different lags. For both signals, the most irregularity was observed during the high valence, and the least irregularity was seen during the low valence. Consequently, signals' irregularity is affected by the valence dimension. The results showed that the fusion has more potential for emotion recognition than that of using each signal separately. For sigma = 0.1, the highest recognition rate was 100% in a subject-dependent mode. In a subject-independent mode, the maximum accuracies of 88.57 and 86.8% were obtained for arousal and valence dimensions, respectively.

Evaluation of Non-Invasive Measurement of Haemoglobin using PPG in Clinically Ill Pediatric Patients

Non-invasive haemoglobin (SpHb) estimation using Photoplethysmograph signal has gained enormous attention among researches in order to provide an earlydiagnosis to polycythemia, anaemia, various cardiovascular diseases, etc. The primary objective of this work is to evaluate the performance efficiency of SpHb monitoring using PPG in clinically ill pediatric population. PPG signal was obtained from the pediatric patients, and SpHb was calculated from the characteristic features of PPG. Haemoglobin value obtained through venous blood sample was compared with SpHb. Theabsolute mean difference between the SpHb and Hbref was 0.78g/dL (SD 0.99; 0.1 to 4.1).For a statistical analysis of the correlation between SpHb and Hblab, IBM SPSS statistics software was used. Bland-Altman analysis, T-test and Linear regression analysis were further used for finding the agreeability limits. Overestimation of SpHb value was observed for lower Hblab values, and SpHb failed to detect anaemic subjects.

P801Continuous cuff-less blood pressure monitoring using vascular elasticity model on photoplethysmogram

Abstract Background Hypertension is a well-known risk factor for potentially cardiovascular diseases, which is underdiagnosis and treatment. Clinical evidence suggests that periodic and continuous monitoring of blood pressure (BP) is helpful for the early detection of hypertension to reduce the cardiovascular risk. Photoplethysmogram (PPG) is a simple and non-invasive technique to measure the instantaneous blood flow in capillaries. This study proposes a miniaturized PPG sensor, which is feasible to integrate into smartphone, to estimate cuff-less BP and heart rate (HR). Methods A miniaturized PPG sensor, provided by FocalTech Smart Sensors Co., Ltd., was used to collect reflective finger PPG signal. The methodology of BP estimation from PPG was using a vascular elasticity model which estimating the physiological parameters from PPG features Simultaneously, we utilized oscillometric blood pressure monitors to obtain automated BP readings as the validation standard. Result Seventy-one subjects were recruited in this validation study. By using our PPG algorithm, an accuracy of ± 11.9 mmHg for systolic BP and ± 8.0 mmHg for diastolicBP was observed at 71 subjects. There was a good linear correlation (CC) between PPG-estimating BP and automated BP (systolic BP, CC=0.4; diastolic BP, CC=0/.45).Likewise, slopes of the liner fit on the Bland Altman plots for both systolic BP (slope = −0.23) and diastolic BP (slope = −0.11) indicated there was no particular bias in predicting high or low BP. Subjects characteristics Parameters Total number (subjects) 71 Entry BP range Normal 21 Prehypertension 27 Stage 1 hypertension 17 Stage 2 hypertension 6 Sex Male 37 Female 34 Age (years) Mean ± standard deviation 55±12 Range 27–77 Conclusion In this study, the feasibility of mobile based cuff-less BP measurement using a vascular elasticity model on finger PPG signal was demonstrated with a level of acceptable accuracy. Based on mobile-based and easy-to-use features of our sensor/device, the incorporation this real-time measurement into the monitoring of BP could be helpful for the management of hypertension.

Photoplethysmogram-based Cognitive Load Assessment Using Multi-Feature Fusion Model

Cognitive load assessment is crucial for user studies and human--computer interaction designs. As a noninvasive and easy-to-use category of measures, current photoplethysmogram- (PPG) based assessment methods rely on single or small-scale predefined features to recognize responses induced by people’s cognitive load, which are not stable in assessment accuracy. In this study, we propose a machine-learning method by using 46 kinds of PPG features together to improve the measurement accuracy for cognitive load. We test the method on 16 participants through the classical n-back tasks (0-back, 1-back, and 2-back). The accuracy of the machine-learning method in differentiating different levels of cognitive loads induced by task difficulties can reach 100% in 0-back vs. 2-back tasks, which outperformed the traditional HRV-based and single-PPG-feature-based methods by 12--55%. When using “leave-one-participant-out” subject-independent cross validation, 87.5% binary classification accuracy was reached, which is at the state-of-the-art level. The proposed method can also support real-time cognitive load assessment by beat-to-beat classifications with better performance than the traditional single-feature-based real-time evaluation method.

In obstructive sleep apnea patients, automatic determination of respiratory arrests by photoplethysmography signal and heart rate variability.

Obstructive sleep apnea is a disease that occurs in connection to pauses in respiration during sleep. Detection of the disease is achieved using a polysomnography device, which is the gold standard in diagnosis. Diagnosis is made by the steps of sleep staging and respiration scoring. Respiration scoring is performed with at least four signals. Technical knowledge is required for attaching the electrodes. Additionally, the electrodes are disturbing to an extent that will delay the patient's sleep. It is needed to have systems as alternatives for polysomnography devices that will bring a solution to these issues. This study proposes a new approach for the process of respiration scoring which is one of the diagnostic steps for the disease. A machine-learning-based apnea detection algorithm was developed for the process of respiration scoring. The study used Photoplethysmography (PPG) signal and Heart Rate Variability (HRV) that is derived from this signal. The PPG records obtained from the patient and control groups were cleaned out using a digital filter. Then, the HRV parameter was derived from the PPG signal. Later, 46 features were derived from the PPG signal and 40 features were derived from the HRV. The derived features were classified with reduced machine-learning techniques using the F-score feature-selection algorithm. The evaluation was made in a multifaceted manner. Besides, Principal Component Analysis was performed to reduce system input (features). According to the results, if a real-time embedded system is designed, the system can operate with 16 PPG feature 95%, four PPG feature 93.81% accuracy rate. These success rates are highly sufficient for the system to work. Considering all these values, it is possible to realize a practical respiration scoring system. With this study, it was agreed upon that PPG signal may be used in the diagnosis of this disease by processing it with machine learning and signal processing techniques.

Sleep-Wake Classification using Statistical Features Extracted from Photoplethysmographic Signals.

Sleep quality has a significant impact on human mental and physical health. Detecting sleep-wake stages is of paramount importance in the study of sleep. The gold standard method for sleep-wake stages classification is the multi-sensors based polysomnography (PSG) systems, which is normally recorded in clinical settings. The main drawback of PSG is the inconvenience to the subjects and can hamper the normal sleep. This paper describes an automated approach for classifying sleep-wake stages using finger-tip photoplethysmographic (PPG) signal. The proposed system used statistical features of PPG signal and supervised machine learning models including K-nearest neighbors (KNN) and support vector machine (SVM). The models are trained using 80% events (3486 sleep-wake events) from the dataset and the rest 20% events (872 sleep-wake events) are used for testing. On the test events, cubic KNN, weighted KNN, quadratic SVM and medium Gaussian SVM show 69.27%, 70.53%, 71.33% and 72.36% overall accuracy respectively for predicting the sleep and wake stages. This result advocates that the statistical features of PPG are capable of recognizing the changes in physiological states. The KNN and SVM classifier adopt the statistical features from PPG signal to differentiate between the wake and sleep stages.

Cuff-less Blood Pressure Measurement Using Supplementary ECG and PPG Features Extracted Through Wavelet Transformation.

Cuff-less blood pressure (BP) is essential for continuous health monitoring to prevent diseases such as hypertension. Due to the discomfort caused by inflation and deflation of the cuff, it is not possible to monitor continuously. Although pulse transit time (PTT) based approach is commonly used, other parameters also vary with BP. Multi-parameter models are developed using regression analysis, to estimate systolic blood pressure (SBP) and diastolic blood pressure (DBP). Hence, the correlation of multiple extracted features with the blood pressure is proved. To achieve this, simultaneous electrocardiogram (ECG) and photoplethysmographic (PPG) along with respective BP data were collected. The developed algorithm uses wavelet transformation on ECG and PPG signals for detection of the occurrence of essential wave points precisely, even in the presence of artifacts. Pulse wave analysis (PWA) is performed to create a feature vector. From the experimental results, it is found that, the SBP model gives mean error of 0.4916 mmHg, with standard deviation of 6.3986 mmHg, whereas for DBP model, mean error is 0.2527 mmHg and standard deviation is 3.2835 mmHg which is acceptable as per the British Hypertension Society (BHS) and Association for the Advancement of Medical Instruments (AAMI) standards. After the removal of BP oddities, mean absolute error improves from 5.625 to 3.854 mmHg for SBP and from 2.564 to 2.144 mmHg for DBP.

Fabrication of a portable device for stress monitoring using wearable sensors and soft computing algorithms

Stress is an issue that everyone experiences in today’s modern life. Prolonged exposure to stress can cause many mental and physical diseases. Accordingly, the stress management issue has become popular, and the need for personal healthcare devices has increased in recent years. Therefore, the aim of this research is to design and manufacture a portable stress monitoring system, based on photoplethysmography (PPG) and galvanic skin response (GSR) physiological signals, acquired by wearable sensors. To do so, we proposed a novel algorithm for continuous measurement of the stress index (SI) as well as the classification of stress levels. In order to estimate an accurate value for SI, various soft computing algorithms such as support vector regression, artificial neural networks (ANN), and adaptive neuro-fuzzy inference system (ANFIS) were adopted for modeling the stress based on the features extracted from normalized and non-normalized types of PPG and GSR signals and their combinations. Furthermore, K-nearest neighbor (KNN), ANNs, Naive Bayes, and support vector machine (SVM) were utilized to discriminate different levels of stress in subjects. The obtained results indicate that the ANFIS algorithm can estimate the SI training output with the correlation coefficient (CC) of 0.9281 and the average relative error of 0.23 on a subset of the combined features of PPG and GSR signals. Also, the best classification performance was for KNN (K = 3) algorithm, with 85.3% accuracy. To evaluate the developed system, data of 16 subjects, out of the training dataset, participated in the experiment in the presence of the experts and psychologists, were used. The average CC of 0.81 and classification accuracy of 75% were obtained, using the implemented ANFIS model and KNN classifier.

An Automated Algorithm to Extract Time Plane Features from the PPG Signal and its Derivatives for Personal Health Monitoring Application

Recently, photoplethysmogram (PPG) signal is widely adopted in health monitoring devices for automated assessment of different cardiovascular parameters. However, research in the area of computerized health analysis using PPG signal features is still lagging behind. In this paper, a robust, automated yet simple algorithm is proposed for accurate detection of characteristic points from the PPG signal and its derivatives. The methodology follows amplitude thresholding, slope-reversal, and an empirical formula based approach. Finally, Baseline modulation is removed from the PPG dataset and features are extracted from the amplitude normalized PPG signal and its derivatives. Performance of the proposed algorithm is evaluated over MIMIC database as well as over real PPG data acquired from both healthy volunteers and cardiac patients. The algorithm exhibits high efficiency for all detected fiducial points with an average sensitivity, positive predictivity and detection accuracy of 99.80%, 99.84%, and 99.65%, respectively. Compared to the existing methods, the proposed algorithm offers complete characterization of the PPG signal and its derivatives.

Learning-Based Screening of Endothelial Dysfunction From Photoplethysmographic Signals

Endothelial-Dysfunction (ED) screening is of primary importance to early diagnosis cardiovascular diseases. Recently, approaches to ED screening are focusing more and more on photoplethysmography (PPG)-signal analysis, which is performed in a threshold-sensitive way and may not be suitable for tackling the high variability of PPG signals. The goal of this work was to present an innovative machine-learning (ML) approach to ED screening that could tackle such variability. Two research hypotheses guided this work: (H1) ML can support ED screening by classifying PPG features; and (H2) classification performance can be improved when including also anthropometric features. To investigate H1 and H2, a new dataset was built from 59 subject. The dataset is balanced in terms of subjects with and without ED. Support vector machine (SVM), random forest (RF) and k-nearest neighbors (KNN) classifiers were investigated for feature classification. With the leave-one-out evaluation protocol, the best classification results for H1 were obtained with SVM (accuracy = 71%, recall = 59%). When testing H2, the recall was further improved to 67%. Such results are a promising step for developing a novel and intelligent PPG device to assist clinicians in performing large scale and low cost ED screening.

A Deep Neural Network-Based Pain Classifier Using a Photoplethysmography Signal.

Side effects occur when excessive or low doses of analgesics are administered compared to the required amount to mediate the pain induced during surgery. It is important to accurately assess the pain level of the patient during surgery. We proposed a pain classifier based on a deep belief network (DBN) using photoplethysmography (PPG). Our DBN learned about a complex nonlinear relationship between extracted PPG features and pain status based on the numeric rating scale (NRS). A bagging ensemble model was used to improve classification performance. The DBN classifier showed better classification results than multilayer perceptron neural network (MLPNN) and support vector machine (SVM) models. In addition, the classification performance was improved when the selective bagging model was applied compared with the use of each single model classifier. The pain classifier based on DBN using a selective bagging model can be helpful in developing a pain classification system.

Hypertension Assessment Using Photoplethysmography: A Risk Stratification Approach.

Hypertension is a common chronic cardiovascular disease (CVD). Early screening and diagnosis of hypertension plays a major role in its prevention and in the control of CVDs. Our study discusses the early screening of hypertension while using the morphological features of photoplethysmography (PPG). Numerous morphological features of PPG and its derivative waves were defined and extracted. Six types of feature selection methods were chosen to screen and evaluate these PPG morphological features. The optimal features were comprehensively analyzed in relation to the physiological processes of the cardiovascular circulatory system. Particularly, the intrinsic relation and physiological significance between the formation process of systolic blood pressure (SBP) and PPG morphology features were analyzed in depth. A variety of linear and nonlinear classification models were established for the comparison trials. The F1 scores for the normotension versus prehypertension, normotension and prehypertension versus hypertension, and normotension versus hypertension trials were 72.97%, 81.82%, and 92.31%, respectively. In summary, this study established a PPG characteristic analysis model and established the intrinsic relationship between SBP and PPG characteristics. Finally, the risk stratification of hypertension at different stages was examined and compared based on the optimal feature subset.

Bidirectional Recurrent Auto-Encoder for Photoplethysmogram Denoising.

Photoplethysmography (PPG) has become ubiquitous with the development of smart watches and the mobile healthcare market. However, PPG is vulnerable to various types of noises that are ever present in uncontrolled environments, and the key to obtaining meaningful signals depends on successful denoising of PPG. In this context, algorithms have been developed to denoise PPG, but many were validated in controlled settings or are reliant on multiple steps that must all work correctly. This paper proposes a novel PPG denoising algorithm based on bidirectional recurrent denoising auto-encoder (BRDAE) that requires minimal pre-processing steps and have the benefit of waveform feature accentuation beyond simple denoising. The BRDAE was trained and validated on a dataset with artificially augmented noise, and was tested on a large open database of PPG signals collected from patients enrolled in intensive care units as well as from PPG data collected intermittently during the daily routine of nine subjects over 24h. Denoising with the trained BRDAE improved signal-to-noise ratio of the noise-augmented data by 7.9dB during validation. In the test datasets, the denoised PPG showed statistically significant improvement in heart rate detection as compared with the original PPG in terms of correlation to reference and root-mean-squared error. These results indicate that the proposed method is an effective solution for denoising the PPG signal, and promises values beyond traditional denoising by providing PPG feature accentuation for pulse waveform analysis.

Human emotion recognition using deep belief network architecture

Recently, deep learning methodologies have become popular to analyse physiological signals in multiple modalities via hierarchical architectures for human emotion recognition. In most of the state-of-the-arts of human emotion recognition, deep learning for emotion classification was used. However, deep learning is mostly effective for deep feature extraction. Therefore, in this research, we applied unsupervised deep belief network (DBN) for depth level feature extraction from fused observations of Electro-Dermal Activity (EDA), Photoplethysmogram (PPG) and Zygomaticus Electromyography (zEMG) sensors signals. Afterwards, the DBN produced features are combined with statistical features of EDA, PPG and zEMG to prepare a feature-fusion vector. The prepared feature vector is then used to classify five basic emotions namely Happy, Relaxed, Disgust, Sad and Neutral. As the emotion classes are not linearly separable from the feature-fusion vector, the Fine Gaussian Support Vector Machine (FGSVM) is used with radial basis function kernel for non-linear classification of human emotions. Our experiments on a public multimodal physiological signal dataset show that the DBN, and FGSVM based model significantly increases the accuracy of emotion recognition rate as compared to the existing state-of-the-art emotion classification techniques.

Eliminating Individual Bias to Improve Stress Detection from Multimodal Physiological Data.

Stress monitoring is important for mental wellbeing and early detection of related disorders. The current work is focused on stress detection from multiple non-invasive physiological signals like Electroencephalogram (EEG), Photoplethysmogram (PPG) and Galvanic Skin Response (GSR). We show that, compared to using only the well known EEG band powers in different frequencies for stress detection, an early fusion with GSR and PPG features shows a significant improvement. Maximum Relevance Minimum Redundancy (mRMR) based feature selection is used to identify the most suitable physiological features correlating with stress. A major contribution of this work lies in eliminating subject-specific bias to improve the classification accuracy. We use self-reported values of Valence, Arousal and Dominance to cluster subjects and build separate classification models specific to clusters. The proposed approach is validated on a publicly available dataset comprising 146 data instances from 10 subjects. The performances of Leave-One- Subject-Out cross validation (LOSOCV) in terms of mean Fscores are 0.61 using EEG features only, 0.64 using early fusion of EEG, GSR and PPG features and 0.69 by applying our clustering technique before fusion and classification.