Epoch Length Research Articles

P100 How Various Epoch Lengths and Activity Cut-Points Influence Physical Activity Estimates for Low-Income Hispanic Adults Under Free-Living Conditions

Background Accelerometers are popular, objective physical activity (PA) measures. However, when using accelerometry, data processing decisions must be made regarding sampling intervals (i.e., epoch lengths (EL)) and PA intensity cut-points. Objective To examine how different epoch lengths (EL) and intensity cut-points influence PA estimates among free-living, low-income Hispanic adults. Study Design, Settings, Participants This study was cross-sectional. Low-income Hispanic adults (n = 35) recruited from Expanded Food and Nutrition Education Program in five states, were asked to wear an accelerometer (ActiGraph GT3X+) for seven consecutive days. Measurable Outcome/Analysis Only data from participants with > 4 valid days (i.e., > 10 hour of wear time) were analyzed. Downloaded data were reintegrated into 1-, 10-, 30-, and 60-second ELs. Within each EL, PA cut-points were classified as either: a) sedentary; b) light; c) moderate; or d) vigorous, using two widely-accepted algorithms (Freedson and Matthews). Total moderate to vigorous PA (MVPA) was also calculated. Since the data were not normally distributed, Kruskal-Wallis analysis was used to compare the time spent in different PA intensity levels, by aforementioned ELs, for each intensity cut-point. Results The participants were mostly female (88.6%), with a mean age of 36.5 ± 11.0. Using the Freedson cut-points, the mean rank of time spent in light PA was higher for longer ELs (H = 90.7, P P P P P Conclusion This study's findings suggest that PA estimates vary by EL regardless of the cut-points used, thus validity studies should be done to compare accelerometer results with indirect calorimetry to generate the optimal EL for estimating low-income Hispanic adults’ true PA levels. Funding NIFA.

Towards Automatic Artifact Rejection in Resting-State MEG Recordings: Evaluating the Performance of the SOUND Algorithm.

In this study, a new automated noise rejection algorithm, the SOurce-estimate-Utilizing Noise-Discarding algorithm (SOUND), was evaluated on magnetoencephalographic (MEG) resting-state signals in order to select its optimal configuration parameters. Different values of the epoch length and the regularization parameter λ0 were assessed in three scenarios with ascending noise levels. Results show that it is possible to remarkably improve the Signal-to-Noise Ratio, without overly altering the signal of interest. An optimal λ0 value of 0.1 was obtained. However, the epoch length should be adapted to the specific problem. In conclusion, our results suggest that the SOUND algorithm is an appropriate and useful tool to be applied in a preprocessing pipeline for MEG restingstate signals.

Non-Intrusive Detection of Drowsy Driving Based on Eye Tracking Data

Drowsy driving is one of the leading causes of motor vehicle accidents in North America. This paper presents the use of eye tracking data as a non-intrusive measure of driver behavior for detection of drowsiness. Eye tracking data were acquired from 53 subjects in a simulated driving experiment, whereas the simultaneously recorded multichannel electroencephalogram (EEG) signals were used as the baseline. A random forest (RF) and a non-linear support vector machine (SVM) were employed for binary classification of the state of vigilance. Different lengths of eye tracking epoch were selected for feature extraction, and the performance of each classifier was investigated for every epoch length. Results revealed a high accuracy for the RF classifier in the range of 88.37% to 91.18% across all epoch lengths, outperforming the SVM with 77.12% to 82.62% accuracy. A feature analysis approach was presented and top eye tracking features for drowsiness detection were identified. Altogether, this study showed a high correspondence between the extracted eye tracking features and EEG as a physiological measure of vigilance and verified the potential of these features along with a proper classification technique, such as the RF, for non-intrusive long-term assessment of drowsiness in drivers. This research would ultimately lead to development of technologies for real-time assessment of the state of vigilance, providing early warning of fatigue and drowsiness in drivers.

Accelerometer-derived physical activity estimation in preschoolers \u2013 comparison of cut-point sets incorporating the vector magnitude vs the vertical axis

BackgroundActiGraph accelerometers are a widely used tool to objectively measure physical activity (PA) behavior in young children and several validated cut-point sets exist to estimate time spent in different PA intensities (sedentary time, light PA, moderate-to-vigorous PA). Applying different cut-point sets leads to large and meaningful differences in results. So far, only cut-point sets validated for the vertical axis have been compared and only the influence on time spent in moderate-to-vigorous PA has been analyzed.MethodsA range of validated cut-point sets with their respective epoch length was applied to analyze cross-sectional data of the Swiss Preschoolers’ Health Study (SPLASHY): 1) Vertical axis in combination with an epoch length of 15 s (VA-15), 2) Vertical axis in combination with an epoch length of 60 s (VA-60) and 3) Vector magnitude in combination with an epoch length of 60 s (VM-60). PA was measured for eight consecutive days using ActiGraph accelerometers (wGT3X-BT). Three days were required to be included in the analysis (minimum two weekdays and one weekend-day with at least ten hours recording per day).ResultsFour hundred forty-five preschoolers (mean age 3.9 ± 0.5 years; 46% girls) had valid accelerometer measurements. A longer epoch (VA-60 vs VA-15) resulted in 2% less sedentary time (ST), 18% more light PA (LPA) and 51% less moderate-to-vigorous PA (MVPA); using the vector magnitude compared to the vertical axis (VM-60 vs VA-60) resulted in 34% less ST, 27% more LPA and 63% more MVPA (all p ≤ 0.001). Comparing all three sets of cut-points, ST ranged from 4.0 to 6.2 h, LPA from 5.1 to 7.6 h and MVPA from 0.8 to 1.6 h.ConclusionsEstimated time spent in different PA intensities was strongly influenced by the choice of cut-point sets. Both, axis selection and epoch length need to be considered when comparing different studies especially when they relate PA behavior to health. The differences in the prevalence of children fulfilling PA guidelines highlight the relevance of these findings.Trial registrationCurrent Controlled Trials ISRCTN41045021 (date of registration: 21.03.2014).

Erratum: Kim, Y., et al. Comparison of Polar Active Watch and Waist- and Wrist-Worn ActiGraph Accelerometers for Measuring Children's Physical Activity Levels during Unstructured Afterschool Programs. Int. J. Environ. Res. Public Health 2018, 15, 2268.

Due to an error during production, the Chandler's cut-points at the 30-sec epoch length [...].

Fractal and Multifractal Properties of Electrographic Recordings of Human Brain Activity: Toward Its Use as a Signal Feature for Machine Learning in Clinical Applications.

The quantification of brain dynamics is essential to its understanding. However, the brain is a system operating on multiple time scales, and characterization of dynamics across time scales remains a challenge. One framework to study such dynamics is that of fractal geometry; and currently there exist several methods for the study of brain dynamics using fractal geometry. We aim to highlight some of the practical challenges of applying fractal geometry to brain dynamics—and as a putative feature for machine learning applications, and propose solutions to enable its wider use in neuroscience. Using intracranially recorded electroencephalogram (EEG) and simulated data, we compared monofractal and multifractal methods with regards to their sensitivity to signal variance. We found that both monofractal and multifractal properties correlate closely with signal variance, thus not being a useful feature of the signal. However, after applying an epoch-wise standardization procedure to the signal, we found that multifractal measures could offer non-redundant information compared to signal variance, power (in different frequency bands) and other established EEG signal measures. We also compared different multifractal estimation methods to each other in terms of reliability, and we found that the Chhabra-Jensen algorithm performed best. Finally, we investigated the impact of sampling frequency and epoch length on the estimation of multifractal properties. Using epileptic seizures as an example event in the EEG, we show that there may be an optimal time scale (i.e., combination of sampling frequency and epoch length) for detecting temporal changes in multifractal properties around seizures. The practical issues we highlighted and our suggested solutions should help in developing robust methods for the application of fractal geometry in EEG signals. Our analyses and observations also aid the theoretical understanding of the multifractal properties of the brain and might provide grounds for new discoveries in the study of brain signals. These could be crucial for the understanding of neurological function and for the developments of new treatments.

A MULTI-LAYER HYBRID MACHINE LEARNING MODEL FOR AUTOMATIC SLEEP STAGE CLASSIFICATION

Sleep Stage Classification (SSC) is a standard process in the Polysomnography (PSG) for studying sleep patterns and events. The SSC provides sleep stage information of a patient throughout an entire sleep test. A physician uses results from SSCs to diagnose sleep disorder symptoms. However, the SSC data processing is time-consuming and requires trained sleep technicians to complete the task. Over the years, researchers attempted to find alternative methods, which are known as Automatic Sleep Stage Classification (ASSC), to perform the task faster and more efficiently. Proposed ASSC techniques usually derived from existing statistical methods and machine learning (ML) techniques. The objective of this study is to develop a new hybrid ASSC technique, Multi-Layer Hybrid Machine Learning Model (MLHM), for classifying sleep stages. The MLHM blends two baseline ML techniques, Decision Tree (DT) and Support Vector Machine (SVM). It operates on a newly developed multi-layer architecture. The multi-layer architecture consists of three layers for classifying [Formula: see text], [Formula: see text] and [Formula: see text], [Formula: see text], [Formula: see text] in different epoch lengths. Our experiment design compares MLHM and baseline ML techniques and other research works. The dataset used in this study was derived from the ISRUC-Sleep database comprising of 100 subjects. The classification performances were thoroughly reviewed using the hold-out and the 10-fold cross-validation method in both subject-specific and subject-independent classifications. The MLHM achieved a certain satisfactory classification results. It gained 0.694[Formula: see text][Formula: see text][Formula: see text]0.22 of accuracy ([Formula: see text]) in subject-specific classification and 0.942[Formula: see text][Formula: see text][Formula: see text]0.02 of accuracy ([Formula: see text]) in subject-independent classification. The pros and cons of the MLHM with the multi-layer architecture were thoroughly discussed. The effect of class imbalance was rationally discussed towards the classification results.

Application and Validation of Activity Monitors' Epoch Lengths and Placement Sites for Physical Activity Assessment in Exergaming.

We assessed the agreement of two ActiGraph activity monitors (wGT3X vs. GT9X) placed at the hip and the wrist and determined an appropriate epoch length for physical activity levels in an exergaming setting. Forty-seven young adults played a 30-min exergame while wearing wGT3X and GT9X on both hip and wrist placement sites and a heart rate sensor below the chest. Intraclass correlation coefficient indicated that intermonitor agreement in steps and activity counts was excellent on the hip and good on the wrist. Bland-Altman plots indicated good intermonitor agreement in the steps and activity counts on both placement sites but a significant intermonitor difference was detected in steps on the wrist. Time spent in sedentary and physical activity intensity levels varied across six epoch lengths and depended on the placement sites, whereas time spent from a 1-s epoch of the hip-worn monitors most accurately matched the relative exercise intensity by heart rate. Hip placement site was associated with better step-counting accuracy for both activity monitors and more valid estimation of physical activity levels. A 1-s epoch was the most appropriate epoch length to detect short bursts of intense physical activity and may be the best choice for data processing and analysis in exergaming studies examining intermittent physical activities.

The effect of different time epoch settings on the classification of sheep behaviour using tri-axial accelerometry

Monitoring behaviour of grazing animals is important for the management of grazing systems. A study was run to discriminate between the main behaviours (grazing, ruminating and other activities) of sheep at pasture wearing a halter equipped with an accelerometer (BEHARUM device), and to identify the epoch setting (5, 10, 30, 60, 120, 180 and 300 s) with the best performance. The BEHARUM device includes a three-axial accelerometer sensor and a force sensor positioned under the lower jaw of the animal. The halter was fitted to eight Sarda dairy sheep that rotationally grazed either a spatial association (mixture) or a time association of berseem clover (Trifolium alexandrinum L.) and Italian ryegrass (Lolium multiflorum Lam.) for 6 h day−1. The behaviour of the animals was also video-recorded. The raw acceleration and force data were processed for each epoch setting to create 15 variables: the mean, variance and inverse coefficient of variation (ICV; mean/standard deviation) per minute for the X-, Y-, Z-axis and force, and the resultant. Multivariate statistical techniques were used to discriminate between the three behavioural activities: canonical discriminant analysis (CDA), and discriminant analysis (DA). To validate the derived discriminant functions, a bootstrap procedure was run. To evaluate the performance of DA in discriminating between the three activities, the sensitivity, specificity, precision, accuracy and Coehn’s k coefficient were calculated, based on the error distribution in assignment. Results show that a discriminant analysis can accurately classify important behaviours such as grazing, ruminating and other activities in sheep at pasture. The prediction model has demonstrated a better performance in classifying grazing behaviour than ruminating and other activities for all epochs. The 30 s epoch length yielded the most accurate classification in terms of accuracy and Coehn’s k coefficient. Nevertheless, 60 and 120 s may increase the potential recording time without causing serious lack of accuracy.

Children's physical activity and sedentary time compared using assessments of accelerometry counts and muscle activity level.

BackgroundThis research compared accelerometry (ACC)-derived and muscle electromyography (EMG)-based estimates of physical activity (PA) and sedentary time in typical PA tasks and during the daily lives of children.MethodsData was included from two exploratory studies. In Study I, 6–7-year-old children (n = 11, 64% girls) were assessed for eight PA tasks (walking, stair negotiation, climbing, crawling, swinging, balancing, trampoline jumping and a game of tag). In Study II, 7–9-year-old children (n = 14, 38% girls) were assessed for six PA tasks (walking, sitting, static squat, single leg hops, jump for height and standing long jump), and daily PA during one day with and one day without structured exercise. Quadriceps and hamstring muscle activity and inactivity using EMG shorts and acceleration by waist-mounted accelerometer were simultaneously measured and classified as sedentary, light, moderate and vigorous activity. Data from ACC was further analyzed using five different published cut-off points and varying time windows (1−60 s) for comparison with EMG.ResultsIn the PA tasks ACC counts and EMG amplitude showed marked differences in swinging, trampoline jumping, crawling, static squat, single leg hops, standing long jump and jump for height, the difference being over 170% when signals were normalized to that during walking. Furthermore, in walking, swinging, trampoline jumping, stair negotiation and crawling ACC classified over 60% of the time as vigorous-intensity activity, while EMG indicated primarily light- and moderate-intensity activities. During both days with and without exercise, ACC resulted in greater proportion of light activity (p < 0.01) and smaller proportion of moderate activity compared to EMG (p < 0.05). The choice of cut-off points and epoch length in ACC analysis influenced the classification of PA level and sedentary time. In the analysis of daily activities the cut-off points by Evenson et al. (2008) with epochs of 7.5 s and 15 s yielded the smallest difference (less than 10% of recording time at each intensity) against EMG-derived PA levels.DiscussionThis research provides novel insight on muscle activity and thereby on neuromuscular loading of major locomotor muscles during normal daily activities of children. While EMG and ACC provided similar estimates of sedentary time in 13 typical PA tasks, duration of light, moderate and vigorous PA varied considerably between the methods especially during walking, stair negotiation, crawling, swinging and trampoline jumping. Evenson et al.’s (2008) cut-off points with ≤15 s epoch provided similar classification of PA than EMG during daily life. Compared to impacts recorded using ACC, EMG can provide understanding on children’s neuromuscular loading during motor tasks that is useful when studying effects of PA interventions on, and development of, motor competence and coordination.

P30. A novel application of the Phase-lag-Index in functional connectivity research

The phase-lag-index (PLI) ( Stam et al., 2007 ) is a commonly used method to quantify functional connectivity (FC) in EEG/MEG data. When calculated for short epochs in the order of several seconds, PLI varies considerably from epoch to epoch. Using resting-state EEG data from 105 healthy subjects, we demonstrate that the pattern of correlations between PLI time-series is characteristic for an individual’s brain. In addition to acting as an identifying fingerprint, this pattern is also highly gender-specific and bears characteristics that are shared among people and also between hemispheres. Furthermore, optimal performance is achieved using an epoch length of 250 ms which is in the same range as typical durations of phase-synchronisation ( Varela et al., 2001 ). This observation hints at the timing-relation between phase-synchronisation events being the underlying source of information captured by correlation between PLI time-series. In conclusion, our work reveals a novel way of extracting meaningful information about the brain’s functional organisation from EEG data.

Effect of Epoch Length on Compressed Sensing of EEG.

Aging populations are stretching existing healthcare systems to their limits in both developing and developed countries. Telemedicine is a promising solution to this challenging problem. Under the conventional data compression paradigm, long-time recording of electroencephalography (EEG) signals still generates excessive amount of data, which requires large data storage and long transmission time. While promoting mobile telemedicine with compressed sensing (CS) as a key system for EEG monitoring, this paper investigates the effect of epoch length on CS to compress EEG signals. Experimental results show that a longer epoch length leads to better signal compression at the expense of larger signal reconstruction time. At a sampling frequency of 256 Hz, a 4-s epoch length is suitable when using a general desktop computer to perform signal reconstruction.

Onset Detection of Epileptic Seizures From Accelerometry Signal.

Epileptic seizures are the result of any abnormal asynchronous firing of cortical neurons. Seizures are abrupt and pose a risk of injury and fatal harm to the patient. Epilepsy affects patients quality of life (QOL) and imposes financial, social, and physical burden on the patient. The unpredictability associated with seizures further adds to the reduced QOL and increases dependence on caregivers and family members. A seizure triggered alarm system can reduce the risk of seizure-related injuries and aid in improving patient's QOL. This study presents real-time onset detection of seizures from accelerometry signal. An automated approach based on statistical machine learning is employed to learn the onset of seizures. To search for the optimal parameter that simultaneously maximizes detection sensitivity (sens) while minimizing false alarm rate (FAR) and latency, the epoch length is varied from $t=\{1,~10s\}$. Linear and non-linear time-varying dynamical patterns were extracted from every epoch using Poincaré plot analysis. The correlation patterns were learned using a kernalized support vector data descriptor. The preliminary analysis on accelerometry data collected from 8 epileptic patients with 9 generalized tonicclonic seizures (GTCS) shows promising results. The proposed algorithm detected all GTCS events (sens: 100%, FAR: 1. 09/24h) at 8s from onset. The proposed algorithm can lead to a sensitive, specific, and a relatively short-latency detection system for real-time remote monitoring of epileptic patients.

Qualitative and quantitative comparative study of accelerometer cut-points to evaluate physical activity in children with cerebral palsy

Physical activity (PA) in daily life could be evaluated by accelerometer. Multiple algorithms have been validated to classify activity counts along several time intervals (epoch length), and with several set of cut-points. For children, three algorithms are the most frequently used by researchers in order to define the time spent in sedentary (SED) and in moderate to vigorous physical activity (MVPA). The aim of this study was to determine the more appropriate accelerometer based classification algorithms for children with CP. Participants were children and adolescents with CP (n = 20, age 10–18 y), with Gross Motor Function Classification System (GMFCS) levels I–III. Daily PA was measured with an Actigraph GT3X accelerometer over 7 consecutive days in free-living environment. Data collection protocols were the same for all children. The results of three different algorithms, Puyau (2002), Freedson (2005), Evenson (2008), were compared by ANOVA followed by multiple t-test after Bonferroni's corrections: The results were also compared to qualitative data from the written diaries of daily PA completed by children and parents. There is a significant difference between the results of physical activity, depending on the algorithm used (F = 16.88, P = 1.53 × 10−15). The average time spent in MVPA is 2.6 times higher with Freedson's algorithm than with Puyau's algorithm, whereas the Puyau and Evenson algorithms are comparable. The qualitative informations collected with the daily diary indicated that Freedson's algorithm over-estimated PA among children with CP. The harmonisation of the methods regarding accelerometer's data collection and processing is necessary. Especially to compare the level of PA of a specific patient to the World Health Organization recommendations. For children with CP, the more appropriate algorithm is based on Evenson's cut-points.

IMS-DTM: Incremental Multi-Scale Dynamic Topic Models

Dynamic topic models (DTM) are commonly used for mining latent topics in evolving web corpora. In this paper, we note that a major limitation of the conventional DTM based models is that they assume a predetermined and fixed scale of topics. In reality, however, topics may have varying spans and topics of multiple scales can co-exist in a single web or social media data stream. Therefore, DTMs that assume a fixed epoch length may not be able to effectively capture latent topics and thus negatively affect accuracy. In this paper, we propose a Multi-Scale Dynamic Topic Model (MS-DTM) and a complementary Incremental Multi-Scale Dynamic Topic Model (IMS-DTM) inference method that can be used to capture latent topics and their dynamics simultaneously, at different scales. In this model, topic specific feature distributions are generated based on a multi-scale feature distribution of the previous epochs; moreover, multiple scales of the current epoch are analyzed together through a novel multi-scale incremental Gibbs sampling technique. We show that the proposed model significantly improves efficiency and effectiveness compared to the single scale dynamic DTMs and prior models that consider only multiple scales of the past.

The effect of epoch length on time and frequency domain parameters of electromyographic and mechanomyographic signals

The selection of epoch lengths affects the time and frequency resolution of electromyographic (EMG) and mechanomyographic (MMG) signals, as well as decisions regarding the signal processing techniques to use for determining the power density spectrum. No previous studies, however, have examined the effects of epoch length on parameters of the MMG signal. The purpose of this study was to examine the differences between epoch lengths for EMG amplitude, EMG mean power frequency (MPF), MMG amplitude, and MMG MPF from the VL and VM muscles during MVIC muscle actions as well as at each 10% of the time to exhaustion (TTE) during a continuous isometric muscle action of the leg extensors at 50% of MVIC. During the MVIC trial, there were no significant (p > 0.05) differences between epoch lengths (0.25, 0.50, 1.00, and 2.00-s) for mean absolute values for any of the EMG or MMG parameters. During the submaximal, sustained muscle action, however, absolute MMG amplitude and MMG MPF were affected by the length of epoch. All epoch related differences were eliminated by normalizing the absolute values to MVIC. These findings supported normalizing EMG and MMG parameter values to MVIC and utilizing epoch lengths that ranged from 0.25 to 2.00-s.

Assessing worst case scenarios in movement demands derived from global positioning systems during international rugby union matches: Rolling averages versus fixed length epochs.

The assessment of competitive movement demands in team sports has traditionally relied upon global positioning system (GPS) analyses presented as fixed-time epochs (e.g., 5–40 min). More recently, presenting game data as a rolling average has become prevalent due to concerns over a loss of sampling resolution associated with the windowing of data over fixed periods. Accordingly, this study compared rolling average (ROLL) and fixed-time (FIXED) epochs for quantifying the peak movement demands of international rugby union match-play as a function of playing position. Elite players from three different squads (n = 119) were monitored using 10 Hz GPS during 36 matches played in the 2014–2017 seasons. Players categorised broadly as forwards and backs, and then by positional sub-group (FR: front row, SR: second row, BR: back row, HB: half back, MF: midfield, B3: back three) were monitored during match-play for peak values of high-speed running (>5 m·s-1; HSR) and relative distance covered (m·min-1) over 60–300 s using two types of sample-epoch (ROLL, FIXED). Irrespective of the method used, as the epoch length increased, values for the intensity of running actions decreased (e.g., For the backs using the ROLL method, distance covered decreased from 177.4 ± 20.6 m·min-1 in the 60 s epoch to 107.5 ± 13.3 m·min-1 for the 300 s epoch). For the team as a whole, and irrespective of position, estimates of fixed effects indicated significant between-method differences across all time-points for both relative distance covered and HSR. Movement demands were underestimated consistently by FIXED versus ROLL with differences being most pronounced using 60 s epochs (95% CI HSR: -6.05 to -4.70 m·min-1, 95% CI distance: -18.45 to -16.43 m·min-1). For all HSR time epochs except one, all backs groups increased more (p < 0.01) from FIXED to ROLL than the forward groups. Linear mixed modelling of ROLL data highlighted that for HSR (except 60 s epoch), SR was the only group not significantly different to FR. For relative distance covered all other position groups were greater than the FR (p < 0.05). The FIXED method underestimated both relative distance (~11%) and HSR values (up to ~20%) compared to the ROLL method. These differences were exaggerated for the HSR variable in the backs position who covered the greatest HSR distance; highlighting important consideration for those implementing the FIXED method of analysis. The data provides coaches with a worst-case scenario reference on the running demands required for periods of 60–300 s in length. This information offers novel insight into game demands and can be used to inform the design of training games to increase specificity of preparation for the most demanding phases of matches.

Description of nighttime cough epochs in patients with stable COPD GOLD II&amp;ndash;IV

PurposeChronic cough is one of the main symptoms of COPD. Ambulatory objective monitoring provides novel insights into the determinants and characteristics of nighttime cough in COPD.Materials and methodsNighttime cough was monitored objectively by LEOSound lung sound monitor in patients with stable COPD II–IV. In 30 patients, with 10 patients in each stage group, nighttime cough was analyzed for epoch frequency, epoch severity (epoch length and coughs per epoch), and pattern (productive or nonproductive).ResultsCough was found in all patients ranging from 1 to 294 events over the recording period. In 29 patients, cough epochs were monitored, ranging from 1 to 75 epochs. The highest amount of cough epochs was found in patients with COPD stage III. Active smokers had significantly more productive cough epochs (61%) than nonsmokers (24%).ConclusionWe found a high rate of nighttime cough epochs in patients with COPD, especially in those in stage III. Productive cough was predominantly found in patients with persistent smoking. LEOSound lung sound monitor offers a practical and valuable opportunity to evaluate cough objectively.

A novel quick seizure detection and localization through brain data mining on ECoG dataset

Epilepsy is a common neurological disorder, and epileptic seizure detection is a scientific challenge since sometimes patient do not experience any alert. The objective of this research is to reduce the seizure detection time while maintaining high accuracy, and locate the brain hemisphere that is mostly affected by seizure. We argue that by using a decision forest (i.e., an ensemble of carefully built decision trees), instead of a single classifier such as a decision tree, we can afford to reduce epoch lengths (used for converting the ECoG and EEG signal into datasets) without compromising accuracy. This will allow us to build the future records in a shorter time resulting in a quicker seizure detection. In this paper, we apply two decision forest classifiers, called SysFor and Forest CERN, on an ECoG brain dataset. Our initial experiments on the dataset of a single patient indicate that decision forest algorithms such as SysFor and Forest CERN can reduce the seizure detection time significantly while maintaining 100% accuracy. They can also be used to identify the region of the brain of a patient that is mostly affected by seizure.

A novel, fast and efficient single-sensor automatic sleep-stage classification based on complementary cross-frequency coupling estimates

ObjectiveLimitations of the manual scoring of polysomnograms, which include data from electroencephalogram (EEG), electro-oculogram (EOG), electrocardiogram (ECG) and electromyogram (EMG) channels have long been recognized. Manual staging is resource intensive and time consuming, and thus considerable effort must be spent to ensure inter-rater reliability. As a result, there is a great interest in techniques based on signal processing and machine learning for a completely Automatic Sleep Stage Classification (ASSC). MethodsIn this paper, we present a single-EEG-sensor ASSC technique based on the dynamic reconfiguration of different aspects of cross-frequency coupling (CFC) estimated between predefined frequency pairs over 5 s epoch lengths. The proposed analytic scheme is demonstrated using the PhysioNet Sleep European Data Format (EDF) Database with repeat recordings from 20 healthy young adults. We validate our methodology in a second sleep dataset. ResultsWe achieved very high classification sensitivity, specificity and accuracy of 96.2 ± 2.2%, 94.2 ± 2.3%, and 94.4 ± 2.2% across 20 folds, respectively, and also a high mean F1 score (92%, range 90–94%) when a multi-class Naive Bayes classifier was applied. High classification performance has been achieved also in the second sleep dataset. ConclusionsOur method outperformed the accuracy of previous studies not only on different datasets but also on the same database. SignificanceSingle-sensor ASSC makes the entire methodology appropriate for longitudinal monitoring using wearable EEG in real-world and laboratory-oriented environments.