Artificial intelligence-enhanced epileptic seizure detection by wearables.

OBJECTIVE: Wrist- or ankle-worn devices are less intrusive than the widely used electroencephalographic (EEG) systems for monitoring epileptic seizures. Using custom-developed deep-learning seizure detection models, we demonstrate the detection of a broad range of seizure types by wearable signals. METHODS: Patients admitted to the epilepsy monitoring unit were enrolled and asked to wear wearable sensors on either wrists or ankles. We collected patients' electrodermal activity, accelerometry (ACC), and photoplethysmography, from which blood volume pulse (BVP) is derived. Board-certified epileptologists determined seizure onset, offset, and types using video and EEG recordings per the International League Against Epilepsy 2017 classification. We applied three neural network models-a convolutional neural network (CNN) and a CNN-long short-term memory (LSTM)-based generalized detection model and an autoencoder-based personalized detection model-to the raw time-series sensor data to detect seizures and utilized performance measures, including sensitivity, false positive rate (the number of false alarms divided by the total number of nonseizure segments), number of false alarms per day, and detection delay. We applied a 10-fold patientwise cross-validation scheme to the multisignal biosensor data and evaluated model performance on 28 seizure types. RESULTS: We analyzed 166 patients (47.6% female, median age = 10.0 years) and 900 seizures (13 254 h of sensor data) for 28 seizure types. With a CNN-LSTM-based seizure detection model, ACC, BVP, and their fusion performed better than chance; ACC and BVP data fusion reached the best detection performance of 83.9% sensitivity and 35.3% false positive rate. Nineteen of 28 seizure types could be detected by at least one data modality with area under receiver operating characteristic curve > .8 performance. SIGNIFICANCE: Results from this in-hospital study contribute to a paradigm shift in epilepsy care that entails noninvasive seizure detection, provides time-sensitive and accurate data on additional clinical seizure types, and proposes a novel combination of an out-of-the-box monitoring algorithm with an individualized person-oriented seizure detection approach.

Predictive Modeling of Secondary Pulmonary Hypertension in Left Ventricular Diastolic Dysfunction.

Diastolic dysfunction is a common pathology occurring in about one third of patients affected by heart failure. This condition may not be associated with a marked decrease in cardiac output or systemic pressure and therefore is more difficult to diagnose than its systolic counterpart. Compromised relaxation or increased stiffness of the left ventricle induces an increase in the upstream pulmonary pressures, and is classified as secondary or group II pulmonary hypertension (2018 Nice classification). This may result in an increase in the right ventricular afterload leading to right ventricular failure. Elevated pulmonary pressures are therefore an important clinical indicator of diastolic heart failure (sometimes referred to as heart failure with preserved ejection fraction, HFpEF), showing significant correlation with associated mortality. However, accurate measurements of this quantity are typically obtained through invasive catheterization and after the onset of symptoms. In this study, we use the hemodynamic consistency of a differential-algebraic circulation model to predict pulmonary pressures in adult patients from other, possibly non-invasive, clinical data. We investigate several aspects of the problem, including the ability of model outputs to represent a sufficiently wide pathologic spectrum, the identifiability of the model's parameters, and the accuracy of the predicted pulmonary pressures. We also find that a classifier using the assimilated model parameters as features is free from the problem of missing data and is able to detect pulmonary hypertension with sufficiently high accuracy. For a cohort of 82 patients suffering from various degrees of heart failure severity, we show that systolic, diastolic, and wedge pulmonary pressures can be estimated on average within 8, 6, and 6 mmHg, respectively. We also show that, in general, increased data availability leads to improved predictions.

Coherence between two coupled lasers from a dynamics perspective.

We compare a simple dynamical model of fiber laser arrays with independent experiments on two coupled lasers. The degree of agreement with experimental observations is excellent. Collectively the evidence presented supports this dynamical approach as an alternative to the traditional static eigenmode analysis of the coupled laser cavities.

Refined fiber laser model.

With modification, a recently proposed laser array model is found to agree quantitatively with fiber laser experiments. Comparisons of transient behavior, stable dynamical states, and transitions are made using both previously published and new experiments. While the original model agrees well for fibers with relatively low losses, achieving quantitative agreement over a wide range of operating conditions requires more physically appropriate descriptions of gain dynamics. The refined model is derived, and its predictions are found to be in excellent agreement with experiments.

Robust synchronization in fiber laser arrays.

Synchronization of coupled fiber lasers has been reported in recent experiments [Bruesselbach, Opt. Lett. 30, 1339 (2005); Minden, Proc. SPIE 5335, 89 (2004)]. While these results may lead to dramatic advances in laser technology, the mechanism by which these lasers synchronize is not understood. We analyze a recently proposed [Rogers, IEEE J. Quantum Electron. 41, 767 (2005)] iterated map model of fiber laser arrays to explore this phenomenon. In particular, we look at synchronous solutions of the maps when the gain fields are constant. Determining the stability of these solutions is analytically tractable for a number of different coupling schemes. We find that in the most symmetric physical configurations the most symmetric solution is either unstable or stable over insufficient parameter range to be practical. In contrast, a lower symmetry configuration yields surprisingly robust coherence. This coherence persists beyond the pumping threshold for which the gain fields become time dependent.

Complex-ordered patterns in shaken convection.

We report and analyze complex patterns observed in a combination of two standard pattern forming experiments. These exotic states are composed of two distinct spatial scales, each displaying a different temporal dependence. The system is a fluid layer experiencing forcing from both a vertical temperature difference and vertical time-periodic oscillations. Depending on the parameters these forcing mechanisms produce fluid motion with either a harmonic or a subharmonic temporal response. Over a parameter range where these mechanisms have comparable influence the spatial scales associated with both responses are found to coexist, resulting in complex, yet highly ordered patterns. Phase diagrams of this region are reported and criteria to define the patterns as quasiperiodic crystals or superlattices are presented. These complex patterns are found to satisfy four-mode (resonant tetrad) conditions. The qualitative difference between the present formation mechanism and the resonant triads ubiquitously used to explain complex-ordered patterns in other nonequilibrium systems is discussed. The only exception to quantitative agreement between our analysis based on Boussinesq equations and laboratory investigations is found to be the result of breaking spatial symmetry in a small parameter region near onset.

Vortices and the entrainment transition in the two-dimensional Kuramoto model.

We study synchronization in the two-dimensional lattice of coupled phase oscillators with random intrinsic frequencies. When the coupling K is larger than a threshold K{E} , there is a macroscopic cluster of frequency-synchronized oscillators. We explain why the macroscopic cluster disappears at K{E} . We view the system in terms of vortices, since cluster boundaries are delineated by the motion of these topological defects. In the entrained phase (K>K{E}) , vortices move in fixed paths around clusters, while in the unentrained phase (K

Renormalization group approach to oscillator synchronization.

We develop a renormalization group method to investigate synchronization clusters in a one-dimensional chain of nearest-neighbor coupled phase oscillators. The method is best suited for chains with strong disorder in the intrinsic frequencies and coupling strengths. The results are compared with numerical simulations of the chain dynamics and good agreement in several characteristics is found. We apply the renormalization group and simulations to Lorentzian distributions of intrinsic frequencies and couplings and investigate the statistics of the resultant cluster sizes and frequencies, as well as the dependence of the characteristic cluster length upon parameters of these Lorentzian distributions.

Universality in the one-dimensional chain of phase-coupled oscillators.

We apply a recently developed renormalization-group (RG) method to study synchronization in a one-dimensional chain of phase-coupled oscillators in the regime of weak randomness. The RG predicts how oscillators with randomly distributed frequencies and couplings form frequency-synchronized clusters. Although the RG was originally intended for strong randomness, i.e., for distributions with long tails, we find good agreement with numerical simulations even in the regime of weak randomness. We use the RG flow to derive how the correlation length scales with the width of the coupling distribution in the limit of large coupling. This leads to the identification of a universality class of distributions with the same critical exponent nu . We also find universal scaling for small coupling. Finally, we show that the RG flow is characterized by a universal approach to the unsynchronized fixed point, which provides physical insight into low-frequency clusters.

Unique facial features distinguish fetal alcohol syndrome patients and controls in diverse ethnic populations.

BACKGROUND: Effective management of fetal alcohol spectrum disorders (FASD) is dependent on the timely and reliable diagnosis of affected individuals. There are significant diagnostic difficulties because of the reduced prominence of facial features as children age to adulthood as well as potential population or ethnic differences in the most characteristic alcohol-related facial features. METHODS: A total of 276 subjects were recruited from 4 sites (Cape Town, South Africa; Helsinki, Finland; Buffalo, New York; and San Diego, California) and completed a detailed dysmorphology evaluation to classify subjects as either fetal alcohol syndrome (FAS; 43%) or control (57%). Computerized anthropometry was employed to identify facial features that could distinguish FAS patients from controls across a wide age range and across ethnically disparate study populations. RESULTS: Subjects were placed into 1 of 4 populations based on their ancestry (Cape Coloured, Finnish Caucasian, African American, or North American Caucasian). Analyses performed in each of the 4 study populations were able to identify a unique set of variables which provided excellent discrimination between the 2 groups (FAS, control). In each study group, at least one ocular-related measurement, shortened palpebral fissure, reduced outer canthal width, or reduced inner canthal width, was included in the final classification model. CONCLUSIONS: We found measurements that reflected reduced size of the eye orbit to be a consistent feature discriminating FAS and controls across each study population. However, each population had a unique, though often overlapping, set of variables which discriminated the 2 groups, suggesting important ethnic differences in the presentation of FAS. It is possible that these differences were accentuated by the wide age distribution of the study subjects.

Self-organized coherence in fiber laser arrays.

We report the production of stable, coherent, and same-phase states in arrays of fiber lasers. Provided that proper interactions between the lasers are present, arrays will spontaneously self-organize into stable coherent same-phase states. There is no need for active control. Power scaling, power spectra, spatial interference fringes, and temporal data all support this conclusion.