16 research outputs found
Sequential levetiracetam and phenytoin in electroencephalographic neonatal seizures unresponsive to phenobarbital: a multicenter prospective observational study in India
Get PDFBackground Although levetiracetam and phenytoin are widely used antiseizure medications (ASM) in neonates, their efficacy fi cacy on seizure freedom is unclear. We evaluated electroencephalographic (EEG) seizure freedom following sequential levetiracetam and phenytoin in neonatal seizures unresponsive to phenobarbital. Methods We recruited neonates born >= 35 weeks and aged <72 h who had continued electrographic seizures despite phenobarbital, from three Indian hospitals, between 20 June 2020 and 31 July 2022. The neonates were treated with intravenous levetiracetam (20 mg/kg x 2 doses, second line) followed by phenytoin (20 mg/kg x 2 doses, third line) if seizures persisted. The primary outcome was complete seizure freedom, defined fi ned as an absence of seizures on EEG for at least 60 min within 40 min from the start of infusion. Findings Of the 206 neonates with continued seizures despite phenobarbital, 152 received levetiracetam with EEG. Of these one EEG was missing, 47 (31.1%) were in status epilepticus, and primary outcome data were available in 145. Seizure freedom occurred in 20 (13.8%; 95% CI 8.6%-20.5%) - 20.5%) after levetiracetam; 16 (80.0%) responded to the fi rst dose and 4 (20.0%) to the second dose. Of the 125 neonates with persisting seizures after levetiracetam, 114 received phenytoin under EEG monitoring. Of these, the primary outcome data were available in 104. Seizure freedom occurred in 59 (56.7%; 95% CI 46.7%-66.4%) - 66.4%) neonates; 54 (91.5%) responded to the fi rst dose and 5 (8.5%) to the second dose. Interpretation With the conventional doses, levetiracetam was associated with immediate EEG seizure cessation in only 14% of phenobarbital unresponsive neonatal seizures. Additional treatment with phenytoin along with levetiracetam attained seizure freedom in further 57%. Safety and efficacy fi cacy of higher doses of levetiracetam should be evaluated in well-designed randomised controlled trials
Generalising diagonal strict concavity property for uniqueness of Nash equilibrium
Get PDFInternational audienceIn this paper, we extend the notion of diagonally strictly concave functions and use it to provide a sufficient condition for uniqueness of Nash equilibrium in some concave games. We then provide an alternative proof of the existence and uniqueness of Nash equilibrium for a network resource allocation game arising from the so-called Kelly mechanism by verifying the new sufficient condition. We then establish that the equilibrium resulting from the differential pricing in the Kelly mechanism is related to a normalised Nash equilibrium of a game with coupled strategy space
Cyclotron instabilities of low frequency, parallel propagating electromagnetic waves in the magnetosphere
Get PDF118-126Electromagnetic electron- and ion-cyclotron instabilities incorporating the details of wave-particle interactions have been studies with reference to low frequency waves in the magnetosphere. The general dispersion relation for transverse electromagnetic waves propagating along the ambient magnetic field in an anisotropic bi-Maxwellian plasma with a mirror loss-cone configuration has been considered. The growth/damping rates for electron-cyclotron waves (whistler) and ion-cyclotron waves have been derived. The electron- and ion-cyclotron wave growths have been computed from the magnetospheric VLF data from ISIS-2 satellite for equatorial and midlatitude auroral regions of the magnetosphere. The dependence of the growth rate of these waves on the temperature anisotropy and mirror loss-cone has been discussed. Loss-cone and electron-cyclotron instabilities are interpreted as the generation mechanism for the low frequency waves in the magnetosphere
Generation mechanism and interpretation of attenuation band of VLF-saucers
Get PDF130-133An analysis of the ISIS-VLF data detected by magnetometer in the auroral region of the magnetosphere has been carried out for VLF-saucers. The generation mechanism of VLF-saucers has been studied in terms of excitation of electrostatic cyclotron harmonic emission due to energetic ions and electrons in the magnetospheric plasma. The attenuation bands of VLF-saucers, as observed by ISIS-2 satellite, have been discussed and it is interpreted that the cyclotron absorption of low energy protons at harmonics of local proton-cyclotron frequencies may be the cause for attenuation bands in VLF-saucers
A Comparative Analysis of In Vitro Toxicity of Synthetic Zeolites on IMR-90 Human Lung Fibroblast Cells
No full textBroad industrial application of zeolites increases the opportunity of inhalation. However, the potential impact of different types and compositions of zeolite on cytotoxicity is still unknown. Four types of synthetic zeolites have been prepared for assessing the effect on lung fibroblast: two zeolite L (LTL-R and LTL-D), ZSM-5 (MFI-S), and faujasite (FAU-S). The cytotoxicity of zeolites on human lung fibroblast (IMR-90) was assessed using WST1 cell proliferation assay, mitochondrial function, membrane leakage of lactate dehydrogenase, reduced glutathione levels, and mitochondrial membrane potential were assessed under control. Intracellular changes were examined using transmission electron microscopy (TEM). Toxicity-related gene expressions were evaluated by PCR array. The result showed significantly higher toxicity in IMR-90 cells with FAU-S than LTL-R, LTL-D and MFI-S exposure. TEM showed FAU-S, spheroidal zeolite with a low Si/Al ratio, was readily internalized forming numerous phagosomes in IMR-90 cells, while the largest and disc-shaped zeolites showed the lowest toxicity and were located in submembranous phagosomes in IMR-90 cells. Differential expression of TNF related genes was detected using PCR arrays and confirmed using qRT-PCR analysis of selected genes. Collectively, the exposure of different zeolites shows different toxicity on IMR-90 cells
User Response Based Recommendations: A Local Angle Approach
Get PDFInternational audienceWhen a user interested in a service/item, visits an online web-portal, it provides description of its interest through initial search keywords. The system recommends items based on these keywords. The user is satisfied if it finds the item of its choice and the system benefits, otherwise the user explores an item from the list. Usually when the user explores an item, it picks an item that is nearest to its interest from the list. While the user explores an item, the system recommends new list of items. This continues till either the user finds its interest or quits. In all, the user provides ample chances and feedback for the system to learn its interest. The aim of this paper is to exploit the user-generated responses in the same session. One can further utilize the history (e.g., previous user ratings) to design good recommendation policies. We develop algorithms that efficiently utilize user responses to recommended items and find the item of user's interest quickly. We first derive optimal policies in the continuous Euclidean space and adapt the same to the space of discrete items. In the continuous Euclidean space, the optimal recommendations (e.g., with two recommendations) at the same time step are at 180 degrees from each other, while are at 90 degrees with respect to the ones at the previous time step. We propose the notion of local angle in the space of discrete items and develop user response-local angle (UR-LA) based recommendation policies. We compared the performance of UR-LA with widely used collaborative filtering (CF) based policies on two real datasets and showed that UR-LA performs better in majority of the test cases. We also proposed a hybrid scheme that combines the best features of both UR-LA and CF (and history) based policies, which outperforms them in most of the cases
SplitEE: Early Exit in Deep Neural Networks with Split Computing
Get PDFDeep Neural Networks (DNNs) have drawn attention because of their outstanding
performance on various tasks. However, deploying full-fledged DNNs in
resource-constrained devices (edge, mobile, IoT) is difficult due to their
large size. To overcome the issue, various approaches are considered, like
offloading part of the computation to the cloud for final inference (split
computing) or performing the inference at an intermediary layer without passing
through all layers (early exits). In this work, we propose combining both
approaches by using early exits in split computing. In our approach, we decide
up to what depth of DNNs computation to perform on the device (splitting layer)
and whether a sample can exit from this layer or need to be offloaded. The
decisions are based on a weighted combination of accuracy, computational, and
communication costs. We develop an algorithm named SplitEE to learn an optimal
policy. Since pre-trained DNNs are often deployed in new domains where the
ground truths may be unavailable and samples arrive in a streaming fashion,
SplitEE works in an online and unsupervised setup. We extensively perform
experiments on five different datasets. SplitEE achieves a significant cost
reduction () with a slight drop in accuracy () as compared to the
case when all samples are inferred at the final layer. The anonymized source
code is available at
\url{https://anonymous.4open.science/r/SplitEE_M-B989/README.md}.Comment: 10 pages, to appear in the proceeding AIMLSystems 202
User Response Based Recommendations: A Local Angle Approach
No full textInternational audienceWhen a user interested in a service/item, visits an online web-portal, it provides description of its interest through initial search keywords. The system recommends items based on these keywords. The user is satisfied if it finds the item of its choice and the system benefits, otherwise the user explores an item from the list. Usually when the user explores an item, it picks an item that is nearest to its interest from the list. While the user explores an item, the system recommends new list of items. This continues till either the user finds its interest or quits. In all, the user provides ample chances and feedback for the system to learn its interest. The aim of this paper is to exploit the user-generated responses in the same session. One can further utilize the history (e.g., previous user ratings) to design good recommendation policies. We develop algorithms that efficiently utilize user responses to recommended items and find the item of user's interest quickly. We first derive optimal policies in the continuous Euclidean space and adapt the same to the space of discrete items. In the continuous Euclidean space, the optimal recommendations (e.g., with two recommendations) at the same time step are at 180 degrees from each other, while are at 90 degrees with respect to the ones at the previous time step. We propose the notion of local angle in the space of discrete items and develop user response-local angle (UR-LA) based recommendation policies. We compared the performance of UR-LA with widely used collaborative filtering (CF) based policies on two real datasets and showed that UR-LA performs better in majority of the test cases. We also proposed a hybrid scheme that combines the best features of both UR-LA and CF (and history) based policies, which outperforms them in most of the cases
