Ahsan H. Khandoker

Support Vector Machines for Automated Recognition of Obstructive Sleep Apnea Syndrome From ECG Recordings

Obstructive sleep apnea syndrome (OSAS) is associated with cardiovascular morbidity as well as excessive daytime sleepiness and poor quality of life. In this study, we apply a machine learning technique [support vector machines (SVMs)] for automated recognition of OSAS types from their nocturnal ECG recordings. A total of 125 sets of nocturnal ECG recordings acquired from normal subjects (OSAS- ) and subjects with OSAS (OSAS+), each of approximately 8 h in duration, were analyzed. Features extracted from successive wavelet coefficient levels after wavelet decomposition of signals due to heart rate variability (HRV) from RR intervals and ECG-derived respiration (EDR) from R waves of QRS amplitudes were used as inputs to the SVMs to recognize OSAS +/- subjects. Using leave-one-out technique, the maximum accuracy of classification for 83 training sets was found to be 100% for SVMs using a subset of selected combination of HRV and EDR features. Independent test results on 42 subjects showed that it correctly recognized 24 out of 26 OSAS + subjects and 15 out of 16 OSAS - subjects (accuracy = 92.85%; Cohens kappa value of 0.85). For estimating the relative severity of OSAS, the posterior probabilities of SVM outputs were calculated and compared with respective apnea/hypopnea index. These results suggest superior performance of SVMs in OSAS recognition supported by wavelet-based features of ECG. The results demonstrate considerable potential in applying SVMs in an ECG-based screening device that can aid a sleep specialist in the initial assessment of patients with suspected OSAS.

An autonomic cloud environment for hosting ECG data analysis services

Advances in sensor technology, personal mobile devices, wireless broadband communications, and Cloud computing are enabling real-time collection and dissemination of personal health data to patients and health-care professionals anytime and from anywhere. Personal mobile devices, such as PDAs and mobile phones, are becoming more powerful in terms of processing capabilities and information management and play a major role in peoples daily lives. This technological advancement has led us to design a real-time health monitoring and analysis system that is Scalable and Economical for people who require frequent monitoring of their health. In this paper, we focus on the design aspects of an autonomic Cloud environment that collects peoples health data and disseminates them to a Cloud-based information repository and facilitates analysis on the data using software services hosted in the Cloud. To evaluate the software design we have developed a prototype system that we use as an experimental testbed on a specific use case, namely, the collection of electrocardiogram (ECG) data obtained at real-time from volunteers to perform basic ECG beat analysis.

Wavelet-Based Feature Extraction for Support Vector Machines for Screening Balance Impairments in the Elderly

Trip related falls are a prevalent problem in the elderly. Early identification of at-risk gait can help prevent falls and injuries. The main aim of this study was to investigate the effectiveness of a wavelet based multiscale analysis of a gait variable [minimum foot clearance (MFC)] in comparison to MFC histogram plot analysis in extracting features for developing a model using support vector machines (SVMs) for screening of balance impairments in the elderly. MFC during walking on a treadmill was recorded on 13 healthy elderly and 10 elderly with a history of tripping falls. Features extracted from MFC histogram and then multiscale exponents between successive wavelet coefficient levels after wavelet decomposition of MFC series were used as inputs to the SVM to classify two gait patterns. The maximum accuracy of classification was found to be 100% for a SVM using a subset of selected wavelet based features, compared to 86.95% accuracy using statistical features. For estimating the relative risk of falls, the posterior probabilities of SVM outputs were calculated. These results suggest superior performance of SVM in the detection of balance impairments based on wavelet-based features and it could also be useful for evaluating for falls prevention intervention.

Association of cardiac autonomic neuropathy with alteration of sympatho-vagal balance through heart rate variability analysis.

Early sub-clinical assessment of severity of cardiac autonomic neuropathy (CAN) and intervention are of prime importance for risk stratification and early treatment in preventing sudden death due to silent myocardial infarction. The Ewing battery is currently the diagnostic tool of choice but is unable to detect sub-clinical disease and requires patient cooperation. Time and frequency domain analysis have several shortcomings including sensitivity to recording length, respiratory activity and non-stationarities in the ECG signal. An important step forward is to have a non-invasive method of detecting CAN that is robust against these shortcomings and has a higher sensitivity for the presence of both sub-clinical and overt clinical disease. This study presents a novel parameter, tone-entropy (T-E) that analyses heart rate variability (HRV) of 20 min lead II ECG recordings. Tone (T) represents sympatho-vagal balance and entropy (E) the autonomic regularity activity. Thirteen normal subjects without (CAN-) and 21 with CAN (CAN+) participated in this study. Among 21 CAN+ subjects, 13 are early CAN+ (eCAN+), eight are definite CAN+ (dCAN+) according to autonomic nervous system function tests as described by Ewing. The results showed that tone was higher and the entropy was lower in the dCAN+ group (T=-0.033 to -0.010 and E=1.73-2.24) compared with the eCAN+ (T=-0.0927 to -0.0311 and E=2.0-2.65) and normal (T=-0.128 to -0.0635 and E=2.64-3.15) group. The research verified that T-E is a suitable method to determine the presence of CAN that correctly identifies experimentally induced changes in cardiac function akin to parasympathetic and sympathetic dysfunction and differentiates between stages in CAN disease progression identified using the Ewing battery.

Automated recognition of patients with obstructive sleep apnoea using wavelet-based features of electrocardiogram recordings

Patients with obstructive sleep apnoea syndrome (OSAS) are at increased risk of developing hypertension and other cardiovascular diseases. This paper explores the use of support vector machines (SVMs) for automated recognition of patients with OSAS types (+/-) using features extracted from nocturnal ECG recordings, and compares its performance with other classifiers. Features extracted from wavelet decomposition of heart rate variability (HRV) and ECG-derived respiration (EDR) signals of whole records (30 learning sets from physionet) are presented as inputs to train the SVM classifier to recognize OSAS+/- subjects. The optimal SVM parameter set is then determined by using a leave-one-out procedure. Independent test results have shown that an SVM using a subset of a selected combination of HRV and EDR features correctly recognized 30/30 of physionet test sets. In comparison, classification performance of K-nearest neighbour, probabilistic neural network, and linear discriminant classifiers on test data was lower. These results, therefore, demonstrate considerable potential in applying SVM in ECG-based screening and can aid sleep specialists in the initial assessment of patients with suspected OSAS.

A comparative study on approximate entropy measure and poincaré plot indexes of minimum foot clearance variability in the elderly during walking

BackgroundTrip-related falls which is a major problem in the elderly population, might be linked to declines in the balance control function due to ageing. Minimum foot clearance (MFC) which provides a more sensitive measure of the motor function of the locomotor system, has been identified as a potential gait parameter associated with trip-related falls in older population. This paper proposes nonlinear indexes (approximate entropy (ApEn) and Poincaré plot indexes) of MFC variability and investigates the relationship of MFC with derived indexes of elderly gait patterns. The main aim is to find MFC variability indexes that well correlate with balance impairments.MethodsMFC data during treadmill walking for 14 healthy elderly and 10 elderly participants with balance problems and a history of falls (falls risk) were analysed using a PEAK-2D motion analysis system. ApEn and Poincaré plot indexes of all MFC data sets were calculated and compared.ResultsSignificant relationships of mean MFC with Poincaré plot indexes (SD1, SD2) and ApEn (r = 0.70, p < 0.05; r = 0.86, p < 0.01; r = 0.74, p < 0.05) were found in the falls-risk elderly group. On the other hand, such relationships were absent in the healthy elderly group. In contrast, the ApEn values of MFC data series were significantly (p < 0.05) correlated with Poincaré plot indexes of MFC in the healthy elderly group, whereas correlations were absent in the falls-risk group. The ApEn values in the falls-risk group (mean ApEn = 0.18 ± 0.03) was significantly (p < 0.05) higher than that in the healthy group (mean ApEn = 0.13 ± 0.13). The higher ApEn values in the falls-risk group might indicate increased irregularities and randomness in their gait patterns and an indication of loss of gait control mechanism. ApEn values of randomly shuffled MFC data of falls risk subjects did not show any significant relationship with mean MFC.ConclusionResults have implication for quantifying gait dynamics in normal and pathological conditions, thus could be useful for the early diagnosis of at-risk gait. Further research should provide important information on whether falls prevention intervention can improve the gait performance of falls risk elderly by monitoring the change in MFC variability indexes.

Sensitivity of temporal heart rate variability in Poincaré plot to changes in parasympathetic nervous system activity

BackgroundA novel descriptor (Complex Correlation Measure (CCM)) for measuring the variability in the temporal structure of Poincaré plot has been developed to characterize or distinguish between Poincaré plots with similar shapes.MethodsThis study was designed to assess the changes in temporal structure of the Poincaré plot using CCM during atropine infusion, 70° head-up tilt and scopolamine administration in healthy human subjects. CCM quantifies the point-to-point variation of the signal rather than gross description of the Poincaré plot. The physiological relevance of CCM was demonstrated by comparing the changes in CCM values with autonomic perturbation during all phases of the experiment. The sensitivities of short term variability (SD 1), long term variability (SD 2) and variability in temporal structure (CCM) were analyzed by changing the temporal structure by shuffling the sequences of points of the Poincaré plot. Surrogate analysis was used to show CCM as a measure of changes in temporal structure rather than random noise and sensitivity of CCM with changes in parasympathetic activity.ResultsCCM was found to be most sensitive to changes in temporal structure of the Poincaré plot as compared to SD 1 and SD 2. The values of all descriptors decreased with decrease in parasympathetic activity during atropine infusion and 70° head-up tilt phase. In contrast, values of all descriptors increased with increase in parasympathetic activity during scopolamine administration.ConclusionsThe concordant reduction and enhancement in CCM values with parasympathetic activity indicates that the temporal variability of Poincaré plot is modulated by the parasympathetic activity which correlates with changes in CCM values. CCM is more sensitive than SD 1 and SD 2 to changes of parasympathetic activity.

Low-Power ECG-Based Processor for Predicting Ventricular Arrhythmia

This paper presents the design of a fully integrated electrocardiogram (ECG) signal processor (ESP) for the prediction of ventricular arrhythmia using a unique set of ECG features and a naive Bayes classifier. Real-time and adaptive techniques for the detection and the delineation of the P-QRS-T waves were investigated to extract the fiducial points. Those techniques are robust to any variations in the ECG signal with high sensitivity and precision. Two databases of the heart signal recordings from the MIT PhysioNet and the American Heart Association were used as a validation set to evaluate the performance of the processor. Based on application-specified integrated circuit (ASIC) simulation results, the overall classification accuracy was found to be 86% on the out-of-sample validation data with 3-s window size. The architecture of the proposed ESP was implemented using 65-nm CMOS process. It occupied 0.112-mm2 area and consumed 2.78-μW power at an operating frequency of 10 kHz and from an operating voltage of 1 V. It is worth mentioning that the proposed ESP is the first ASIC implementation of an ECG-based processor that is used for the prediction of ventricular arrhythmia up to 3 h before the onset.

Comparison of pulse rate variability with heart rate variability during obstructive sleep apnea

We investigate whether pulse rate variability (PRV) extracted from finger photo-plethysmography (Pleth) waveforms can be the substitute of heart rate variability (HRV) from RR intervals of ECG signals during obstructive sleep apnea (OSA). Simultaneous measurements (ECG and Pleth) were taken from 29 healthy subjects during normal (undisturbed sleep) breathing and 22 patients with OSA during OSA events. Highly significant (p<0.01) correlations (1.0>r>0.95) were found between heart rate (HR) and pulse rate (PR). Bland-Altman plot of HR and PR shows good agreement (<5% difference). Comparison of 2 min recording epochs demonstrated significant differences (p<0.01) in time, frequency domains and complexity analysis, between normal and OSA events using PRV as well as HRV measures. Results suggest that both HRV and PRV indices could be used to distinguish OSA events from normal breathing during sleep. However, several variability measures (SDNN, RMSSD, HF power, LF/HF and sample entropy) of PR and HR were found to be significantly (p<0.01) different during OSA events. Therefore, we conclude that PRV provides accurate inter-pulse variability to measure heart rate variability under normal breathing in sleep but does not precisely reflect HRV in sleep disordered breathing.

Toe clearance and velocity profiles of young and elderly during walking on sloped surfaces

BackgroundMost falls in older adults are reported during locomotion and tripping has been identified as a major cause of falls. Challenging environments (e.g., walking on slopes) are potential interventions for maintaining balance and gait skills. The aims of this study were: 1) to investigate whether or not distributions of two important gait variables [minimum toe clearance (MTC) and foot velocity at MTC (VelMTC)] and locomotor control strategies are altered during walking on sloped surfaces, and 2) if altered, are they maintained at two groups (young and elderly female groups).MethodsMTC and VelMTC data during walking on a treadmill at sloped surfaces (+3°, 0° and -3°) were analysed for 9 young (Y) and 8 elderly (E) female subjects.ResultsMTC distributions were found to be positively skewed whereas VelMTC distributions were negatively skewed for both groups on all slopes. Median MTC values increased (Y = 33%, E = 7%) at negative slope but decreased (Y = 25%, E = 15%) while walking on the positive slope surface compared to their MTC values at the flat surface (0°). Analysis of VelMTC distributions also indicated significantly (p < 0.05) lower minimum and 25th percentile (Q1) values in the elderly at all slopes.ConclusionThe young displayed a strong positive correlation between MTC median changes and IQR (interquartile range) changes due to walking on both slopes; however, such correlation was weak in the older adults suggesting differences in control strategies being employed to minimize the risk of tripping.