Search

BACKGROUND AND OBJECTIVES: Seizures (SZs) and other SZ-like patterns of brain activity can harm the brain and contribute to in-hospital death, particularly when prolonged. However, experts qualified to interpret EEG data are scarce. Prior attempts to automate this task have been limited by small or inadequately labeled samples and have not convincingly demonstrated generalizable expert-level performance. There exists a critical unmet need for an automated method to classify SZs and other SZ-like events with expert-level reliability. This study was conducted to develop and validate a computer algorithm that matches the reliability and accuracy of experts in identifying SZs and SZ-like events, known as “ictal-interictal-injury continuum” (IIIC) patterns on EEG, including SZs, lateralized and generalized periodic discharges (LPD, GPD), and lateralized and generalized rhythmic delta activity (LRDA, GRDA), and in differentiating these patterns from non-IIIC patterns. METHODS: We used 6,095 scalp EEGs from 2,711 patients with and without IIIC events to train a deep neural network, SPaRCNet, to perform IIIC event classification. Independent training and test data sets were generated from 50,697 EEG segments, independently annotated by 20 fellowship-trained neurophysiologists. We assessed whether SPaRCNet performs at or above the sensitivity, specificity, precision, and calibration of fellowship-trained neurophysiologists for identifying IIIC events. Statistical performance was assessed by the calibration index and by the percentage of experts whose operating points were below the model's receiver operating characteristic curves (ROCs) and precision recall curves (PRCs) for the 6 pattern classes. RESULTS: SPaRCNet matches or exceeds most experts in classifying IIIC events based on both calibration and discrimination metrics. For SZ, LPD, GPD, LRDA, GRDA, and “other” classes, SPaRCNet exceeds the following percentages of 20 experts—ROC: 45%, 20%, 50%, 75%, 55%, and 40%; PRC: 50%, 35%, 50%, 90%, 70%, and 45%; and calibration: 95%, 100%, 95%, 100%, 100%, and 80%, respectively. DISCUSSION: SPaRCNet is the first algorithm to match expert performance in detecting SZs and other SZ-like events in a representative sample of EEGs. With further development, SPaRCNet may thus be a valuable tool for an expedited review of EEGs. CLASSIFICATION OF EVIDENCE: This study provides Class II evidence that among patients with epilepsy or critical illness undergoing EEG monitoring, SPaRCNet can differentiate (IIIC) patterns from non-IIIC events and expert neurophysiologists.

BACKGROUND AND OBJECTIVES: The validity of brain monitoring using electroencephalography (EEG), particularly to guide care in patients with acute or critical illness, requires that experts can reliably identify seizures and other potentially harmful rhythmic and periodic brain activity, collectively referred to as “ictal-interictal-injury continuum” (IIIC). Previous interrater reliability (IRR) studies are limited by small samples and selection bias. This study was conducted to assess the reliability of experts in identifying IIIC. METHODS: This prospective analysis included 30 experts with subspecialty clinical neurophysiology training from 18 institutions. Experts independently scored varying numbers of ten-second EEG segments as “seizure (SZ),” “lateralized periodic discharges (LPDs),” “generalized periodic discharges (GPDs),” “lateralized rhythmic delta activity (LRDA),” “generalized rhythmic delta activity (GRDA),” or “other.” EEGs were performed for clinical indications at Massachusetts General Hospital between 2006 and 2020. Primary outcome measures were pairwise IRR (average percent agreement [PA] between pairs of experts) and majority IRR (average PA with group consensus) for each class and beyond chance agreement (κ). Secondary outcomes were calibration of expert scoring to group consensus, and latent trait analysis to investigate contributions of bias and noise to scoring variability. RESULTS: Among 2,711 EEGs, 49% were from women, and the median (IQR) age was 55 (41) years. In total, experts scored 50,697 EEG segments; the median [range] number scored by each expert was 6,287.5 [1,002, 45,267]. Overall pairwise IRR was moderate (PA 52%, κ 42%), and majority IRR was substantial (PA 65%, κ 61%). Noise-bias analysis demonstrated that a single underlying receiver operating curve can account for most variation in experts' false-positive vs true-positive characteristics (median [range] of variance explained ([Image: see text]): 95 [93, 98]%) and for most variation in experts' precision vs sensitivity characteristics ([Image: see text]: 75 [59, 89]%). Thus, variation between experts is mostly attributable not to differences in expertise but rather to variation in decision thresholds. DISCUSSION: Our results provide precise estimates of expert reliability from a large and diverse sample and a parsimonious theory to explain the origin of disagreements between experts. The results also establish a standard for how well an automated IIIC classifier must perform to match experts. CLASSIFICATION OF EVIDENCE: This study provides Class II evidence that an independent expert review reliably identifies ictal-interictal injury continuum patterns on EEG compared with expert consensus.

In recent years there has been renewed interest in the behavior of dislocations in crystals that exhibit strong atomic scale disorder, as typical of compositionally complex single phase alloys. The behavior of dislocations in such crystals has been often studied in the framework of elastic manifold pinning in disordered systems. Here we discuss modifications of this framework that may need to be adapted when dealing with extended dislocations that split into widely separated partials. We demonstrate that the presence of a stacking fault gives rise to an additional stress scale that needs to be compared with the pinning stress of elastic manifold theory to decide whether the partials are pinned individually or the dislocation is pinned as a whole. For the case of weakly interacting partial dislocations, we demonstrate the existence of multiple metastable states at stresses below the depinning threshold and analyze the stress evolution of the stacking fault width during loading. In addition we investigate how geometrical constraints can modulate the dislocation-solute interaction and enhance the pinning stress. We compare our theoretical arguments with results of atomistic and discrete (partial) dislocation dynamics (D(P)DD) simulations.

Epilepsy is the second most popular neurological disorder affecting 65 million people around the world. Seizures are classified into two kinds; focal and generalized ictal activities, reflecting the spread of seizure activity on the brain. Focal seizures start and affect specific regions of the brain, whereas the generalized propagates throughout the brain. Current approaches to developing an automatic seizure detection algorithm do not consider the types of seizures. However, to detect the focal seizures, the locations of onset of seizure must be identified by an expert through inspection of the electroencephalogram (EEG), which is an expensive and time-consuming procedure. Moreover, most proposed methods are patient-specific and cannot be generalized on an unseen patient, limiting the clinical usage of previous studies. This work presents a generalizable seizure detection algorithm by considering different seizure types. After pre-processing data and rejecting artifacts, a deep neural network is used to extract robust representations across seizures and a population. The proposed method includes deep recurrent and convolutional neural networks to capture spatial and temporal information simultaneously. Experiments on the TUH EEG seizure dataset, which contains both generalized and focal seizures, show that the proposed method increases the accuracy over state-of-the-art from 80.72% to 82%, precision from 67.55% to 71.69%, and sensitivity from 80% to 85%.

In this paper, a poultry slaughterhouse wastewater (PSW) was treated in terms of chemical oxygen demand (COD) and color reduction using electro-Fenton (EF) technique under response surface methodology (RSM). The effects of five significant independent variables such as reaction time, pH, H2O2/Fe2+ molar ratio, current density, volume ratio of H2O2/PSW (ml/l) were investigated on the COD and color removal. Experimental data were optimized by Box-Behnken design (BBD) and RSM. The optimum conditions were experimentally found at pH of 4.38, reaction time of 55.60 min, H2O2/Fe2+ molar ratio of 3.73, current density of 74.07 mA/cm2, volume ratio of H2O2/PSW of 1.63 ml/l for 92.37%COD removal and at pH of 3.39, reaction time of 49.22 min, H2O2/Fe2+ molar ratio of 3.62, current density of 67.90 mA/cm2, volume ratio of H2O2/PSW of 1.44 ml/l for 88.06% color removal.

Current approaches to automated sleep staging from the electroencephalogram (EEG) rely on constructing a large labeled training and test corpora by aggregating data from different individuals. However, many of the subjects in the training set may exhibit changes in the EEG that are very different from the subjects in the test set. Training an algorithm on such data without accounting for this diversity can cause underperformance. Moreover, test data may have unexpected sensor misplacement or different instrument noise and spectral responses. This work proposes a novel method to learn relevant individuals based on their similarities effectively. The proposed method embeds all training patients into a shared and robust feature space. Individuals who share strong statistical relationships and are similar based on their EEG signals are clustered in this feature space before being passed to a deep learning framework for classification. Using 994 patient EEGs from the 2018 Physionet Challenge (≈6,561 h of recording), we demonstrate that the clustering approach significantly boosts performance compared to state-of-the-art deep learning approaches. The proposed method improves, on average, a precision score from 0.72 to 0.81, a sensitivity score from 0.74 to 0.82, and a Cohen’s Kappa coefficient from 0.64 to 0.75 under 10-fold cross-validation.

We report results of a multiscale simulation study of multilayer structures consisting of graphene sheets with embedded Pt nanoparticles. Density functional theory is used to understand the energetics of Pt-graphene interfaces and provide reference data for the parameterization of a Pt-graphene interaction potential. Molecular dynamics simulations then provide the conformation and energetics of graphene sheets with embedded Pt nanoparticles of varying density, form and size. These results are interpreted using a continuum mechanical model of sheet deformation, and serve to parameterize a meso-scale Monte Carlo model to investigate the question under which conditions the free volume around the Pt nanoparticles forms a percolating cluster, such that the structures can be used in catalytic applications. We conclude with a discussion of potential applications of such multilayer structures., Comment: 21 pages, 10 figures, 2 tables, one supplementary information file

In this work we use density functional theory (DFT) calculations to benchmark empirical potentials for the interaction between nickel and sp$^2$ bonded carbon nanoparticles. These potentials are then used in order to investigate how Ni decorated or coated carbon nanotubes (CNT) affect the mechanical properties of Al/CNT composites. In particular we look at the pull-out behaviour of pristine as well as Ni-decorated and Ni-coated CNT from an Al matrix. Our result shows that Ni coating may produce an extended interface (interphase) where a significant amount of energy is dissipated during CNT pull-out, leading to a high pull-out force. We also demonstrate that surface decorated CNT may act as efficient nano-crystallization agents and thus provide a novel strengthening mechanism not previously discussed in the literature. We discuss our results in view of promising approaches for engineering CNT-metal interfaces such as to achieve high strength metal-CNT composite., 17 pages, 9 figures, 3 tables

We use atomistic simulation (molecular mechanics and molecular dynamics) to investigate failure of graphene sheets containing randomly distributed vacancies. We investigate the dependency of the failure stress on defect concentration and sheet size and show that our findings are consistent with the Duxbury-Leath-Beale (DLB) theory of mechanical or electric breakdown in random media. The corresponding distribution of failure stresses falls into the Gumbel, rather than the Weibull class of extremal statistics. By comparing molecular mechanics and zero-temperature molecular dynamics simulations we establish the role of kinetic energy in crack propagation and its impact on crack patterns emerging before sheet rupture.