Minutes Of Data Research Articles

Total-body dynamic PET/CT imaging reveals kinetic distribution of [13N]NH3 in normal organs.

To systematically investigate kinetic metrics and metabolic trapping of [13N]NH3 in organs. Eleven participants performed total-body [13N]NH3 dynamic positron emission tomography (PET). Regions of interest were drawn in organs to obtain time-to-activity curves (TACs), which were fitted with an irreversible two-tissue compartment model (2TC) to investigate constant rates K1, k2 and k3, and to calculate Ki. Additionally, one-tissue compartment model using full data (1TCfull) and the first four minutes of data (1TC4min) were fitted to TAC data. K1 and k2 were compared among different models to assess [13N]NH3 trapping in organs. Kinetic rates of [13N]NH3 varied significantly among organs. The mean K1 ranged from 0.049 mL/cm3/min in the muscle to 2.936 mL/cm3/min in the kidney. The k2 and k3 were lowest in the liver (0.001min- 1) and in the pituitary (0.009min- 1), while highest in the kidney (0.587min- 1) and in the liver (0.800min- 1), respectively. The Ki was largest in the myocardium (0.601 ± 0.259 mL/cm3/min) while smallest in the bone marrow (0.028 ± 0.022 mL/cm3/min). Three groups of organs with similar kinetic characteristics were revealed: (1) the thyroid, the lung, the spleen, the pancreas, and the kidney; (2) the liver and the muscle; and (3) the cortex, the white matter, the cerebellum, the pituitary, the parotid, the submandibular gland, the myocardium, the bone, and the bone marrow. Obvious k3 was identified in multiple organs, and significant changes of K1 in multiple organs and k2 in most organs were found between 2TC and 1TCfull, but both K1 and k2 were comparable between 2TC and 1TC4min. The kinetic rates of [13N]NH3 differed among organs with some have obvious 13N-anmmonia trapping. The normal distribution of kinetic metrics of 13N-anmmonia in organs can serve as a reference for its potential use in tumor imaging.

Rapid-IAF: Rapid Identification of Individual Alpha Frequency in EEG Data Using Sequential Bayesian Estimation.

Rapid and robust identification of the individual alpha frequency (IAF) in electroencephalogram (EEG) is an essential factor for successful brain-computer interface (BCI) use. Here we demonstrate an algorithm to determine the IAF from short-term resting-state scalp EEG data. First, we outlined the algorithm to determine IAF from short-term resting scalp EEG data and evaluated its reliability using a large-scale dataset of scalp EEG during motor imagery-based BCI use and independent dataset for generalizability confirmation (N = 147). Next, we characterized the relationship between IAF and responsive frequency band of sensorimotor rhythm, which exhibits prominent event-related desynchronization (SMR-ERD) while attempting unilateral and movement. The proposed sequential Bayesian estimation algorithm (Rapid-IAF) determined IAF from less than 26-second resting EEG data among 95% of participants, indicating a clear advance over the conventional methods, which uses 2-15 minutes of data in previous literatures. We confirmed that the determined IAF corresponded to the frequency of SMR, which exhibits the most prominent event-related desynchronization during BCI use (individual SMR-ERD frequency, ISF). Moreover, intraclass correlation revealed that the estimated IAF was more stable than ISF across sessions, suggesting its reliability and utility for robust BCI use without intermittent recalibration. Conclusion In summary, our method rapidly and reliably determined IAF compared to the conventional method using the spectral power change based on task-related response. The method can be utilized to quick BCI initialization. The demonstration of rapid, task-free parametrization of individual variability of neural responses would be of importance for future BCI systems including neural communication via a cursor, an avatar or robots, and closed-loop neurofeedback training.

Decoding of the speech envelope from EEG using the VLAAI deep neural network

To investigate the processing of speech in the brain, commonly simple linear models are used to establish a relationship between brain signals and speech features. However, these linear models are ill-equipped to model a highly-dynamic, complex non-linear system like the brain, and they often require a substantial amount of subject-specific training data. This work introduces a novel speech decoder architecture: the Very Large Augmented Auditory Inference (VLAAI) network. The VLAAI network outperformed state-of-the-art subject-independent models (median Pearson correlation of 0.19, p < 0.001), yielding an increase over the well-established linear model by 52%. Using ablation techniques, we identified the relative importance of each part of the VLAAI network and found that the non-linear components and output context module influenced model performance the most (10% relative performance increase). Subsequently, the VLAAI network was evaluated on a holdout dataset of 26 subjects and a publicly available unseen dataset to test generalization for unseen subjects and stimuli. No significant difference was found between the default test and the holdout subjects, and between the default test set and the public dataset. The VLAAI network also significantly outperformed all baseline models on the public dataset. We evaluated the effect of training set size by training the VLAAI network on data from 1 up to 80 subjects and evaluated on 26 holdout subjects, revealing a relationship following a hyperbolic tangent function between the number of subjects in the training set and the performance on unseen subjects. Finally, the subject-independent VLAAI network was finetuned for 26 holdout subjects to obtain subject-specific VLAAI models. With 5 minutes of data or more, a significant performance improvement was found, up to 34% (from 0.18 to 0.25 median Pearson correlation) with regards to the subject-independent VLAAI network.

Efficient Estimation of Time-Dependent Brain Functional Connectivity Using Anatomical Connectivity Constraints

There is ongoing interest in the dynamics of resting state brain networks (RSNs) as potential predictors of cognitive and behavioural states. Multivariate Autoregressors (MAR) are used to model regional brain activity as a linear combination of past activity in other regions. The coefficients of the MAR are taken as estimates of effective brain connectivity. However, assumption of stationarity, and the large number of coefficients renders the MAR impractical for estimating brain networks from standard neuroimaging time-series of limited durations. We propose HsMM-MAR-AC, a novel sparse hybrid discrete-continuous model for the efficient estimation of time-dependent effective brain networks from non-stationary brain activity time-series. Discrete quasi-stationary Brain States, and the fast switching between them, are modelled by a Hidden semi-Markov Model whose continuous emissions are drawn from a sparse MAR. The coefficients of the MAR are restricted by Anatomical Brain Connectivity information in two ways: 1) Effective direct connectivity between two brain regions is only considered if the corresponding anatomical connection exists; and 2) the autoregressors lag associated with each connection is based on the fiber length between the two regions, such that only one lag per connection is estimated. We test the accuracy of HsMM-MAR-AC in recovering simulated resting state networks of various durations, and at different thresholds of anatomical restrictions. We demonstrate that HsMM-MAR-AC recovers the RSNs more accurately than the benchmark method of the sliding window, with as little as 4 minutes of data. We also show that when the anatomical restrictions are relaxed, longer time-series are needed to estimate the networks, and became computationally unfeasible without anatomical restrictions. HsMM-MAR-AC offers an efficient model for estimating time-dependent Effective Connectivity from neuroimaging data that exploits the advantages of Hidden Markov and MAR models without identifiability problems, excessive demand on data collection, or unnecessary computational effort.

Generalizable electroencephalographic classification of Parkinson's disease using deep learning

Growing interest surrounds the use of electroencephalography (EEG) and deep learning for diagnosing neurological conditions like Parkinson's Disease (PD). Despite the existing proof-of-concept literature demonstrating the potential of deep learning in classifying PD from EEG data, neurologists have been slow to adopt these tools due to insufficient evidence of their real-world diagnostic generalizable performance. Our study aimed to evaluate the potential of deep learning for inter-subject PD classification using a conservative training approach and testing on an external independent dataset. Specifically, we utilized publicly available resting-state EEG data from PD patients at two separate centers, the University of New Mexico (n = 54) and the University of Iowa (n = 28), as our training and testing sets, respectively. Each of these recordings had a minimum of 2 minutes of data. We implemented a channel-wise convolutional neural network, tuning it with a leave-one-subject-out cross-validation approach. Our approach achieved a patient-level accuracy of 80.4% (epoch-level accuracy = 72.7%), which remained consistent when tested on the external dataset (patient-level accuracy = 82.8%, epoch-level accuracy = 75.7%). Our model performs equal-or-better than other standard classification models and our approach compares favourably to similar works. Our publicly available code serves as a foundation for future research exploring different deep learning architectures, investigating other pathologies, and involving larger datasets with the hope of accelerating the adoption of objective computational approaches for the diagnosis and monitoring of neurological disorders.

Assessment of Model Accuracy in Eyes Open and Closed EEG Data: Effect of Data Pre-Processing and Validation Methods.

Eyes open and eyes closed data is often used to validate novel human brain activity classification methods. The cross-validation of models trained on minimally preprocessed data is frequently utilized, regardless of electroencephalography data comprised of data resulting from muscle activity and environmental noise, affecting classification accuracy. Moreover, electroencephalography data of a single subject is often divided into smaller parts, due to limited availability of large datasets. The most frequently used method for model validation is cross-validation, even though the results may be affected by overfitting to the specifics of brain activity of limited subjects. To test the effects of preprocessing and classifier validation on classification accuracy, we tested fourteen classification algorithms implemented in WEKA and MATLAB, tested on comprehensively and simply preprocessed electroencephalography data. Hold-out and cross-validation were used to compare the classification accuracy of eyes open and closed data. The data of 50 subjects, with four minutes of data with eyes closed and open each was used. The algorithms trained on simply preprocessed data were superior to the ones trained on comprehensively preprocessed data in cross-validation testing. The reverse was true when hold-out accuracy was examined. Significant increases in hold-out accuracy were observed if the data of different subjects was not strictly separated between the test and training datasets, showing the presence of overfitting. The results show that comprehensive data preprocessing can be advantageous for subject invariant classification, while higher subject-specific accuracy can be attained with simple preprocessing. Researchers should thus state the final intended use of their classifier.

Traffic state prediction using conditionally Gaussian observed Markov fuzzy switching model

Efficient and accurate prediction of traffic state plays a major role in the implementation of effective intelligent transportation systems. Therefore, traffic forecasting has been attracting a significant interest over the last decades striving to efficiently achieve highest accuracy of reliable prediction algorithms. In this paper, we present a novel prediction algorithm based on the Conditionally Gaussian Observed Markov Fuzzy Switching Model (CGOMFSM). The proposed scheme relies on a triplet representation of traffic encompassing traffic flow, speed and a switch process in order to infer parameters that govern the dynamics of traffic. This work investigates the impact of explicit incorporation of fuzzy switching processes on the accuracy of traffic data prediction. It aims to overcome the shortcomings of several existing prediction schemes in which the crisp modeling of traffic dynamics is hindering the effectiveness of prediction. The experimental study shows that the proposed algorithm yields satisfactory results for a prediction horizon up to 60 minutes for data collected at regular 15-minutes intervals.

Middle‐Aged and Older Adults with Greater White Matter Hyperintensity Burden Display Attenuated Blood Pressure and Cerebral Blood Velocity Responses During Orthostatic Tilt

Cerebral autoregulation (CA) is the process by which the brain maintains adequate blood flow despite sudden changes to perfusion pressure. White matter hyperintensities (WMH) are associated with increased risk of cerebrovascular disease and dementia. This study investigated whether dynamic CA is reduced in cognitively unimpaired middle‐aged and older adults. Additionally, we evaluated dynamic CA and cerebrovascular responses at rest and in response to orthostatic tilt. We hypothesized that CA would be impaired during orthostatic tilt and that diminished blood pressure and cerebral artery blood velocity changes would be associated with greater WMH fraction. We studied 33 middle‐aged and older adults (age = 63 ± 4 y). Intracranial volume (ICV) and WMH lesion volume were measured using a T1 and FLAIR scan, respectively, on a 3T MRI scanner. WMH fraction was calculated as the cube root of (WMH lesion volume/ICV) *100. CA was measured using middle cerebral artery blood velocity (MCAv) from transcranial Doppler ultrasound and beat‐to‐beat mean arterial pressure (MAP) from a finger cuff. Five minutes of data were collected at rest and during 60º orthostatic tilt. Transfer function analysis was used to quantify dynamic CA and reported as gain and phase at low frequency (LF) (0.07‐0.20 Hz). MCAv, MAP and cerebrovascular resistance index (CVRi; calculated as MAP/MCAv) were used to quantify the cerebrovascular response to orthostatic tilt. LF gain was greater at baseline compared to 60º tilt (0.52 cm/s/mmHg vs 0.46 cm/s/mmHg, p &lt; 0.05); however, there were no differences in LF phase between baseline and 60º tilt (29.70 degrees vs 29.42 degrees, p = 0.95). During 60º tilt, MAP and MCAv decreased compared to baseline (MAP: 106 mmHg vs 101 mmHg, p = 0.02; MCAv: 59 cm/s vs 53 cm/s, p &lt; .01). CVRi was higher at 60º tilt compared with baseline (1.90 mmHg/cm/s vs 2.02 mmHg/cm/s, p &lt; 0.01). There were significant inverse associations between the percent change from baseline to 60º tilt in MAP (r = ‐0.4, p = 0.03), MCAv (r = ‐0.39, p = 0.03), but not CVRi (r = 0.13, p = 0.50), and WMH fraction. There were no associations at baseline or 60º tilt between gain (Baseline: r = ‐0.18, p = 0.33; 60º: r = ‐0.31, p = 0.09) or phase (Baseline: r = 0.07, p = 0.71; 60º: r = ‐0.29, p = 0.12) and WMH fraction. Our results indicate that dynamic CA is preserved in cognitively unimpaired adults during tilt. However, individuals with diminished MAP and MCAv responses to orthostatic tilt displayed greater WMH fraction. Future studies are needed to investigate the relationship between blood pressure and cerebral blood flow in individuals with cognitive impairment to further understand the role of cerebral blood flow regulation in the development of cognitive decline.

Analysis and Classification of Stress Among Students Using EEG as Biomarker

As per a report from National Crime Records Bureau (NCRB), in India a total of 10,159 students committed suicide during 2018, which implies that average of 28 students reported suicide every day. This increase in the number that leads to an impact on study and analysis of stress among college students. An experimental research carried out by acquiring EEG data from six students under two different conditions. Before going to write their examination EEG data acquired for 3 minutes with recollection of topics for the examination and after examination 3 minutes of data collected with psychological test carried out for all the eight students using eight channel EEG device. Preprocessing of raw EEG with using band pass filter with corner frequencies of 0.1Hz to 40Hz to eliminate low as well as high frequency signal. EEG features extracted by using discrete wavelet transformation technique with six level of decomposition using db4 as mother wavelet. The main EEG features, such as theta, alpha, and beta relative sub band wave energies, were computed using MATLAB. Four different machine learning algorithms, such as decision trees, support vector machine (SVM), k-Nearest Neighbour (k-NN), and ensemble classifier were used to classify stress or non-stress state. Among four ML classifier algorithm ensemble classifier obtained highest classification accuracy of 87%.

Amputee Fall Risk Classification Using Machine Learning and Smartphone Sensor Data from 2-Minute and 6-Minute Walk Tests.

The 6-min walk test (6MWT) is commonly used to assess a person’s physical mobility and aerobic capacity. However, richer knowledge can be extracted from movement assessments using artificial intelligence (AI) models, such as fall risk status. The 2-min walk test (2MWT) is an alternate assessment for people with reduced mobility who cannot complete the full 6MWT, including some people with lower limb amputations; therefore, this research investigated automated foot strike (FS) detection and fall risk classification using data from a 2MWT. A long short-term memory (LSTM) model was used for automated foot strike detection using retrospective data (n = 80) collected with the Ottawa Hospital Rehabilitation Centre (TOHRC) Walk Test app during a 6-min walk test (6MWT). To identify FS, an LSTM was trained on the entire six minutes of data, then re-trained on the first two minutes of data. The validation set for both models was ground truth FS labels from the first two minutes of data. FS identification with the 6-min model had 99.2% accuracy, 91.7% sensitivity, 99.4% specificity, and 82.7% precision. The 2-min model achieved 98.0% accuracy, 65.0% sensitivity, 99.1% specificity, and 68.6% precision. To classify fall risk, a random forest model was trained on step-based features calculated using manually labeled FS and automated FS identified from the first two minutes of data. Automated FS from the first two minutes of data correctly classified fall risk for 61 of 80 (76.3%) participants; however, <50% of participants who fell within the past six months were correctly classified. This research evaluated a novel method for automated foot strike identification in lower limb amputee populations that can be applied to both 6MWT and 2MWT data to calculate stride parameters. Features calculated using automated FS from two minutes of data could not sufficiently classify fall risk in lower limb amputees.

Accurate Decoding of Imagined and Heard Melodies.

Music perception requires the human brain to process a variety of acoustic and music-related properties. Recent research used encoding models to tease apart and study the various cortical contributors to music perception. To do so, such approaches study temporal response functions that summarise the neural activity over several minutes of data. Here we tested the possibility of assessing the neural processing of individual musical units (bars) with electroencephalography (EEG). We devised a decoding methodology based on a maximum correlation metric across EEG segments (maxCorr) and used it to decode melodies from EEG based on an experiment where professional musicians listened and imagined four Bach melodies multiple times. We demonstrate here that accurate decoding of melodies in single-subjects and at the level of individual musical units is possible, both from EEG signals recorded during listening and imagination. Furthermore, we find that greater decoding accuracies are measured for the maxCorr method than for an envelope reconstruction approach based on backward temporal response functions (bTRFenv). These results indicate that low-frequency neural signals encode information beyond note timing, especially with respect to low-frequency cortical signals below 1 Hz, which are shown to encode pitch-related information. Along with the theoretical implications of these results, we discuss the potential applications of this decoding methodology in the context of novel brain-computer interface solutions.

B-PO03-186 MONITORING ATRIAL FIBRILLATION PATIENTS USING ACTIVE CONTACTLESS VIDEOPLESTHYMOGRAPHY IMPLEMENTED ON A PERSONAL TABLET

Early detection of subclinical AF can help prevent significant complications such as embolic stroke, heart failure, and AF-related hospitalizations. We evaluated the performance of a facial pulsation detection video technology called videoplethysmography (VPG) to detect the presence of AF using a monitoring method that does not require a dedicated sensor and/or cumbersome recording procedures. We enrolled patients with AF undergoing either cardioversion or ablation therapy. Patients were asked to collect two 25-sec video recordings daily using a smart tablet loaded with the VPG technology. Early recurrence of AF during the 15-day follow-up period was based on continuous ECG recording (14-day patch monitor). ECGs were manually annotated for the presence of AF episodes lasting >15-sec in duration. The VPG algorithm measured pulse irregularity and was designed to minimize false positive detections. Development and validation were performed on independent datasets (½ split without a subject in common). We report the specificity, sensitivity, and accuracy of the video-based method to detect early recurrence of AF during the follow-up period. The motion of the patient's face was tracked, but no feedback was provided to the subjects during the measurement. We enrolled 112 patients (age: 65∓9 years, 81 men) over an 18 month period. Five subjects did not collect any quality data while 107 patients collected 7∓5 minutes of data on average at home (min:30sec, max: 25min) over 12∓3 days (min:1.5, max:14.7 days). Thirty subjects (28%) experienced at least one 15-sec AF episode post-procedure. The sensitivity and specificity of the AF detection algorithm based on a total duration of 7 min. of video recordings, over 12 days were 33% and 95% in the validation dataset (accuracy: 74%). Less than one-minute daily monitoring using a contactless VPG method can detect AF in one-third of patients experiencing an early recurrence of AF with a 5% false-positive rate. The sensitivity of the VPG method is expected to increase with a larger number of daily VPG recordings.

Frequency drift in MR spectroscopy at 3T

PurposeHeating of gradient coils and passive shim components is a common cause of instability in the B0 field, especially when gradient intensive sequences are used. The aim of the study was to set a benchmark for typical drift encountered during MR spectroscopy (MRS) to assess the need for real-time field-frequency locking on MRI scanners by comparing field drift data from a large number of sites. MethodA standardized protocol was developed for 80 participating sites using 99 3T MR scanners from 3 major vendors. Phantom water signals were acquired before and after an EPI sequence. The protocol consisted of: minimal preparatory imaging; a short pre-fMRI PRESS; a ten-minute fMRI acquisition; and a long post-fMRI PRESS acquisition. Both pre- and post-fMRI PRESS were non-water suppressed. Real-time frequency stabilization/adjustment was switched off when appropriate. Sixty scanners repeated the protocol for a second dataset. In addition, a three-hour post-fMRI MRS acquisition was performed at one site to observe change of gradient temperature and drift rate. Spectral analysis was performed using MATLAB. Frequency drift in pre-fMRI PRESS data were compared with the first 5:20 minutes and the full 30:00 minutes of data after fMRI. Median (interquartile range) drifts were measured and showed in violin plot. Paired t-tests were performed to compare frequency drift pre- and post-fMRI. A simulated in vivo spectrum was generated using FID-A to visualize the effect of the observed frequency drifts. The simulated spectrum was convolved with the frequency trace for the most extreme cases. Impacts of frequency drifts on NAA and GABA were also simulated as a function of linear drift. Data from the repeated protocol were compared with the corresponding first dataset using Pearson's and intraclass correlation coefficients (ICC). ResultsOf the data collected from 99 scanners, 4 were excluded due to various reasons. Thus, data from 95 scanners were ultimately analyzed. For the first 5:20 min (64 transients), median (interquartile range) drift was 0.44 (1.29) Hz before fMRI and 0.83 (1.29) Hz after. This increased to 3.15 (4.02) Hz for the full 30 min (360 transients) run. Average drift rates were 0.29 Hz/min before fMRI and 0.43 Hz/min after. Paired t-tests indicated that drift increased after fMRI, as expected (p < 0.05). Simulated spectra convolved with the frequency drift showed that the intensity of the NAA singlet was reduced by up to 26%, 44 % and 18% for GE, Philips and Siemens scanners after fMRI, respectively. ICCs indicated good agreement between datasets acquired on separate days. The single site long acquisition showed drift rate was reduced to 0.03 Hz/min approximately three hours after fMRI. DiscussionThis study analyzed frequency drift data from 95 3T MRI scanners. Median levels of drift were relatively low (5-min average under 1 Hz), but the most extreme cases suffered from higher levels of drift. The extent of drift varied across scanners which both linear and nonlinear drifts were observed.

Corticokinematic coherence is stronger to regular than irregular proprioceptive stimulation of the hand.

Proprioceptive afference can be investigated using corticokinematic coherence (CKC), which indicates coupling between limb kinematics and cortical activity. CKC has been quantified using proprioceptive stimulation (movement actuators) with fixed interstimulus interval (ISI). However, it is unclear how regularity of the stimulus sequence (jitter) affects CKC strength. Eighteen healthy volunteers (16 right-handed, 27.8 ± 5.0 yr, 7 females) participated in magnetoencephalography (MEG) session in which their right index finger was continuously moved at ∼3 Hz with Constant 333 ms ISI or with 20% Jitter (ISI 333 ± 66 ms) using a pneumatic-movement actuator. Three minutes of data per condition were collected. Finger kinematics were recorded with a three-axis accelerometer. CKC strength was defined as the peak coherence value in the Rolandic MEG gradiometer pair contralateral to the movement at 3 Hz. Both conditions resulted in significant coherence peaking in the gradiometers over the primary sensorimotor cortex. Constant stimulation yielded stronger CKC at 3 Hz (0.78 ± 0.11 vs. 0.66 ± 0.13, P < 0.001) and its first harmonic (0.60 ± 0.19 vs. 0.27 ± 0.11, P < 0.001) than irregular stimulation. Similarly, the respective sustained-movement evoked field was also stronger for constant stimulation. The results emphasize the importance of temporal stability of the proprioceptive stimulation sequence when quantifying CKC strength. The weaker CKC during irregular stimulation can be explained with temporal and thus spectral scattering of the paired peripheral and cortical events beyond the mean stimulation frequency. This impairs the signal-to-noise ratio of respective MEG signal and thus CKC strength. When accurately estimating and following changes in CKC strength, we suggest using precise movement actuators with constant stimulation sequence.NEW & NOTEWORTHY Cortical proprioceptive processing can be investigated using corticokinematic coherence (CKC). The findings show that CKC method is sensitive to temporal stability in the stimulation sequence. Although both regular and irregular sequences resulted in robust coherence, the regular stimulation sequence with pneumatic movement actuator is recommended to maximize coherence strength and reproducibility to allow better comparability between groups or populations.

Movie-watching outperforms rest for functional connectivity-based prediction of behavior

A major goal of human neuroscience is to relate differences in brain function to differences in behavior across people. Recent work has established that whole-brain functional connectivity patterns are relatively stable within individuals and unique across individuals, and that features of these patterns predict various traits. However, while functional connectivity is most often measured at rest, certain tasks may enhance individual signals and improve sensitivity to behavior differences. Here, we show that compared to the resting state, functional connectivity measured during naturalistic viewing—i.e., movie watching—yields more accurate predictions of trait-like phenotypes in the domains of both cognition and emotion. Traits could be predicted using less than three minutes of data from single video clips, and clips with highly social content gave the most accurate predictions. Results suggest that naturalistic stimuli amplify individual differences in behaviorally relevant brain networks.

Unsupervised hidden semi-Markov model for automatic beat onset detection in 1D Doppler ultrasound

Objective: One dimensional (1D) Doppler ultrasound (DUS) is commonly used for fetal health assessment, during both regular prenatal visits and labor. It is used in preference to ECG and other modalities because of its simplicity and cost. To date, all analysis of such data has been confined to a smoothed, windowed heart rate estimation derived from the 1D DUS signal, reducing the potential of short-term variability information. A first step in improving the assessment of short-term variability of the fetal heart rate (FHR) is through implementing an accurate beat detector for 1D DUS signals. Approach: This work presents an unsupervised probabilistic segmentation method enabled by a hidden semi-Markov model (HSMM). The proposed method employs envelope and spectral features for an online segmentation of fetal 1D DUS signal. The beat onsets and fetal cardiac beat-to-beat intervals are then estimated from the segmentations. For this work, two data sets were used, including 1D DUS recordings from five fetuses recorded in Germany, comprising 6521 beats and 45.06 minutes of data (dataset 1). Simultaneous fetal ECG (fECG) was used as the reference for beat timing. Dataset 2, comprising 4044 beats captured from 17 subjects in the UK was hand scored for beat location and was used as an independent held-out test set. Leave-one-out subject cross-validation was used for parameter tuning on dataset 1. No retraining was performed for dataset 2. To assess the performance of the beat onset detection, the root mean square error (RMSE), F1 score, sensitivity, positive predictivity (PPV) and the error in several standard common heart rate variability metrics were used. These metrics were evaluated on three fiducial points: (1) beat onset, (2) beat offset, and (3) middle of beat interval. Main results: In dataset 1, the proposed method provided an RMSE of 20 ms, F1 score of 97.5 %, a Se of 97.6%, and a PPV of 97.3%. In dataset 2, the proposed method achieved an RMSE of 26 ms, an F1 score of 98.5 %, a Se of 98.0 % and a PPV of 98.9 %. It was also determined that the best beat-to-beat interval was derived from the onset of each beat. For the dataset 2, significant correlations were found in all short term heart rate variability metrics tested, both in the time and frequency domain. Only the proportion of successive normal-to-normal interval differences greater than 20 ms (pNN20) exhibited a significant absolute difference. Significance: This work presents the first-ever description of an algorithm to identify cardiac beats with 1D DUS, closely matching the fetal ECG-derived beats, to enable short-term heart rate variability analysis. The novel algorithm proposed requires no human labeling of data, and could have applicability beyond 1D DUS to other similar highly variable time series.

Dynamic displacements from high-rate GNSS: Error modeling and vibration detection

Nowadays, GNSS (Global Navigation Satellite System) receivers can measure with sampling rates up to 100 Hz, and it is possible to provide displacements at a level of precision of a few millimeters. This opened an interesting field of applications, e.g. for disciplines that monitor dynamic movements, like seismology or structural health monitoring. However, the detection of smallest displacements possible requires a thorough analysis of both error types, systematic and stochastic. In view of this, this paper deals with the detection of small vibrations with amplitudes down to the sub-millimeter level, and the error modelling necessary to achieve this. The data under investigation are time series of kinematic GNSS displacement measurements. We will use parametric time series models of the ARIMA (Autoregressive Integrated Moving Average) type to characterize the GNSS errors and to do a pre-filtering based on a calibrated model. With non-parametric spectral estimation methods – namely periodograms – we can then test for the presence of harmonic processes (sinusoids) in the data. Once a sinusoid is detected, its amplitude and phase are estimated by means of harmonic regression. Subsequently, the sinusoid can be removed, and the procedure can be repeated. This algorithm is tested with data collected for a GNSS shake table experiment; it is shown that it is possible resolve and detect vibrations down to a level of tens of microns with few minutes of data being analyzed. We also highlight some theoretical considerations on minimum detectable amplitudes of vibrations for measurements of different quality and record lengths.

A Time Motion Study Evaluating the Impact of Geographic Cohorting of Hospitalists.

Geographic cohorting (GCh) localizes hospitalists to a unit. Our objective was to compare the GCh and non-GCh workday. In an academic, Midwestern hospital we observed hospitalists in GCh and non-GCh teams. Time in patient rooms was considered direct care; other locations were considered 'indirect' care. Geotracking identified time spent in each location and was obtained for 17 hospitalists. It was supplemented by in-person observation of four GCh and four non-GCh hospitalists for a workday each. Multilevel modeling was used to analyze associations between direct and indirect care time and team and workday characteristics. Geotracking yielded 10,522 direct care episodes. GCh was associated with longer durations of patient visits while increasing patient loads were associated with shorter visits. GCh, increasing patient loads, and increasing numbers of units visited were associated with increased indirect care time. In-person observations yielded 3,032 minutes of data. GCh hospitalists were observed spending 56% of the day in computer interactions vs non-GCh hospitalists (39%; P < .005). The percentage of time spent multitasking was 18% for GCh and 14% for non-GCh hospitalists (P > .05). Interruptions were pervasive, but the highest interruption rate of once every eight minutes in the afternoon was noted in the GCh group. GCh may have the potential to increase patient-hospitalist interactions but these gains may be attenuated if patient loads and the structure of cohorting are suboptimal. The hospitalist workday is cognitively intense. The interruptions noted may increase the time taken for time-intensive tasks like electronic medical record interactions.

DTA-HOC: Online HTTPS traffic service identification using DNS in large-scale networks

An increasing number of websites are making use of HTTPS encryption to enhance security and privacy for their users. However, HTTPS encryption makes it very difficult to identify the service over HTTPS flows, which poses challenges to network security management. In this paper we present DTA-HOC, a novel DNS-based two-level association HTTPS traffic online service identification method for large-scale networks, which correlates HTTPS flows with DNS flows using big data stream processing and association technologies to label the service in an HTTPS flow with a specific associated domain name. DTA-HOC has been specifically designed to address three practical challenges in the service identification process: domain name ambiguity, domain name query invisibility, and data association time window size contradictions. Several experiments on datasets collected from a 10-Gbps campus network are conducted alongside offline and online testing. Results show that DTA-HOC can achieve an average online association rate on HTTPS traffic of 83% and a generic accuracy of 86.16%. Its processing time for one minute of data is less than 20 seconds. These results indicate that DTA-HOC is an efficient method for online identification of services in HTTPS flows for large-scale networks. Moreover, our proposed method can contribute to the identification of other applications which make a Domain Name System (DNS) communication before establishing a connection.

Abstract 16672: Determining the Diagnostic Yield for Atrial Fibrillation Using Continuous Smartwatch Monitoring

Introduction: Smartwatch technologies become ubiquitous in our daily life and can contribute to better health. The optical sensing element (photoplethysmography or PPG) enables continuous heart rate and heart rhythm monitoring. However, signal quality is a major challenge to overcome to improve the diagnostic yield. Hypothesis: To compare the diagnostic yield of AF detection between traditional 24h Holter and continuous wearable monitoring. Methods: In this study, patients receiving a diagnostic 24h ECG Holter monitor were parallel provided with a wearable smartwatch device capable of continuous PPG logging. Data was recorded up to 24 hours and the ECG and PPG signals were synchronized. The PPG signals were analyzed by the FibriCheck algorithms and compared towards the Holter analysis results. Results: In total, 100 patients participated in the study with a mean age of 60±13 years and 53.8% was male. Over 142,560 minutes of data have been collected from both Holter and wearable device. On average, 45% of the data was labeled as insufficient quality. There was a significant (P&lt;0.001) difference in presence of bad quality between day (47.5%) and night (10.6%). In total, on Holter, 9.6% of the data was labelled as atrial fibrillation, while on the PPG signal only 6.5% AF was found. Missing AF occurred during episodes of insufficient quality, impacting the diagnostic yield. Without a quality detection system, a over-diagnostic rate of 48.1% AF was observed. Conclusions: Initial results indicate the feasibility of continuous heart rhythm monitoring using wearable systems. However, optimizing the diagnostic yield remains a challenge. Diagnostic results obtained during the night are in high agreement between Holter and wearable.