Search

Introduction Aims of this study were (a) to assess the development ratio of hyponatremia during treatment with linezolid and (b) to evaluate the relationship between the risk of hyponatremia and linezolid exposure and patient background. Method Clinical data including linezolid serum concentrations and serum sodium values were collected at Toyama University Hospital and Kyorin University Hospital. Data from 89 patients were used for the analysis, and a nadir serum sodium level ≤130 mmol/L during the treatment with linezolid was defined as hyponatremia. Mann-Whitney's U test was used to evaluate the effects of the area under the time-concentration curve (AUC) of linezolid at the nadir sodium level, clinical characteristics (e.g. laboratory data), and baseline serum sodium levels on the development of hyponatremia. Results The hyponatremia was occurred in 21 of 89 patients (23.6%). Data are compared for baseline and nadir serum sodium levels of patients with and without hyponatremia. In both groups, nadir serum sodium levels were significantly different from those of the baseline values (P Conclusions Linezolid exposure, age, and baseline sodium levels were detected as the risk factors for linezolid-related hyponatremia. Our findings suggest that regular monitoring of serum sodium levels is desirable during treatment with linezolid, especially for the elderly and patients with low serum sodium levels before the start of linezolid administration.

Treatment of Candidemia has become increasingly complicated as more and more non-albicans Candida species are being isolated in recent years.We launched an investigation of the species, the MIC value, and the state of administration of antifungal drugs for all the cases with Candida spp. confirmed by blood cultures for the 7-year period from 2012 to 2018 at our hospital. In total, 192 cases were found and 206 strains of Candida species were isolated. Overall, 49.5% of the 206 isolated strains were Candida albicans (102 strains), followed by Candida glabrata (40 strains, 19.4%), and Candida parapsilosis (38 strains, 18.4%). The most frequently used antifungal drug for the initial dose was MCFG (120 cases, 59.2%), while the most frequently switched antifungal agent was L-AMB. Cases with an inappropriate end-of-treatment time represented 58.7% of all the cases.We investigated the Candidemia situation at our hospital for a period of seven years. We believe that it is important for medical institutions to gather detailed data on candidemia at their own hospitals. Likewise, the hospital's Infection Control Team/Antimicrobial Stewardship Team should inform the physicians-in-charge about the appropriate diagnosis and treatment based on the data obtained.

Analogue in-memory computing using memristors could alleviate the performance constraints imposed by digital von Neumann systems in data-intensive tasks. Conventional linear memristors typically operate at high currents, potentially limiting power efficiency and scalability in practical applications. Here, we show that nonlinear ferroelectric tunnel junction memristors can perform linear computation at ultralow currents. Using logarithmic line drivers, we demonstrate that analogue-voltage-amplitude vector–matrix multiplication (VMM) can be performed in selectorless ferroelectric tunnel junction crossbars by exploiting a device nonlinearity factor that remains constant for multiple conductive states. We also show that our ferroelectric tunnel junction crossbars have the attributes required to scale analogue VMM-intensive applications, such as neural inference engines, towards energy efficiencies above 100 tera-operations per second per watt. Nonlinear ferroelectric tunnel junction memristors can be used to perform linear vector–matrix multiplication operations at ultralow currents.

Learning is a fundamental component for creating intelligent machines. Biological intelligence orchestrates synaptic and neuronal learning at multiple time-scales to self-organize populations of neurons for solving complex tasks. Inspired by this, we design and experimentally demonstrate an adaptive hardware architecture Memristive Self-organizing Spiking Recurrent Neural Network (MEMSORN). MEMSORN incorporates resistive memory (RRAM) in its synapses and neurons which configure their state based on Hebbian and Homeostatic plasticity respectively. For the first time, we derive these plasticity rules directly from the statistical measurements of our fabricated RRAM-based neurons and synapses. These “technologically plausible” learning rules exploit the intrinsic variability of the devices and improve the accuracy of the network on a sequence learning task by 30%. Finally, we compare the performance of MEMSORN to a fully-randomly set-up recurrent network on the same task, showing that self-organization improves the accuracy by more than 15%. This work demonstrates the importance of the device-circuit-algorithm co-design approach for implementing brain-inspired computing hardware., Research Square

"Neisseria skkuensis" is a gram-negative coccus that is endemic in the human oral cavity, with only few reports of infection in humans. Herein, we report a case of a male patient in his sixties presenting with infective endocarditis (IE) caused by "N. skkuensis". To our knowledge, this is the second case of IE confirmed using 16S rRNA gene to have been caused by "N. skkuensis". The accurate diagnosis of rare or difficult-to-identify pathogens is a major challenge for clinical microbiological laboratories. Although Neisseria spp. are common in the oral cavity and are often seen in routine tests, identification of their biochemical properties and mass spectrometric analysis are difficult. In this case report, we describe the accurate identification of "N. skkuensis" by 16S rRNA gene sequencing analysis compared to other identification methods. Further cases of "N. skkuensis" are needed to fully evaluate the clinical approach of this detection method.

Spike timing-dependent plasticity (STDP), which is widely studied as a fundamental synaptic update rule for neuromorphic hardware, requires precise control of continuous weights. From the viewpoint of hardware implementation, a simplified update rule is desirable. Although simplified STDP with stochastic binary synapses was proposed previously, we find that it leads to degradation of memory maintenance during learning, which is unfavourable for unsupervised online learning. In this work, we propose a stochastic binary synaptic model where the cumulative probability of the weight change evolves in a sigmoidal fashion with potentiation or depression trials, which can be implemented using a pair of switching devices consisting of serially connected multiple binary memristors. As a benchmark test we perform simulations of unsupervised learning of MNIST images with a two-layer network and show that simplified STDP in combination with this model can outperform conventional rules with continuous weights not only in memory maintenance but also in recognition accuracy. Our method achieves 97.3% in recognition accuracy, which is higher than that reported with standard STDP in the same framework. We also show that the high performance of our learning rule is robust against device-to-device variability of the memristor's probabilistic behaviour.

We study the performance of unsupervised learning using a stochastic synaptic update rule in a spiking neural network with binary synapses. For the binary synapse, we propose a circuit that employs a pair of resistive switching devices with switching properties described by a Weibull distribution. Through simulations, we find that the learning performance is much better with a shape parameter larger than 1, in which case the synaptic operations are regarded as non-Bernoulli stochastic trials, than with a simple Bernoulli model, and that the performance is even better than that achieved using a deterministic rule with continuous synaptic weights.

An analog-to-digital mixed circuit with resistors for static neurons was implemented on a CMOS IC chip. By comparing current magnitudes via the resistors on two lines, the neuron circuit output a multiplier accumulation result and a step function at the comparator as firing. Analog neurons with 1024 synapses and digital peripherals always operated at around 10 mW. The firing delay was intrinsically caused by patterns of inputs and synaptic weights. A recurrent connection directly from the output to the input generated an oscillation, and thus average latency of about 1 μs could be estimated from the observed period. Dynamic firing was observed even with digitally controlled recurrence, indicating a data-converter function from static to dynamic in which the firing pattern can be tuned via randomized synapses. A potential application is reservoir computing, where nonvolatile memristive devices can be further added for readout and learning with timing-dependent plasticity.

Background The objective of this study was to perform an external evaluation of published linezolid population pharmacokinetic and pharmacodynamic models, to evaluate the predictive performance using an independent data set. Another aim was to offer an elegant environment for display and simulation of both the concentration and platelet count after linezolid administration. Methods We performed a systematic literature search in PubMed for all studies evaluating the population pharmacokinetic and pharmacodynamic parameters of linezolid in patients and selected the models to be used for the external validation. The bias of predictions was visually evaluated by plotting prediction errors (PEs) and relative PEs. The precision of prediction was evaluated by calculating the mean absolute error (MAE), root mean squared error (RMSE), and mean relative error (MRE). Results Three articles (models A, B, and C) provided linezolid-induced platelet dynamic models using population pharmacokinetic and pharmacodynamic modeling approaches. The PE and relative PE of both linezolid concentrations and platelet counts for models A and C showed similar predictive distributions. With respect to the prediction accuracy of total linezolid concentration, the MAE, RMSE, and MRE of population prediction values for model C was the smallest. The comparison of the MAE, RMSE, and MRE of patient-individual prediction values for the 3 pharmacodynamic models revealed no large differences. Conclusions We confirmed the transferability of published population pharmacokinetic and pharmacodynamic models and showed that they were suitable for extrapolation to other hospitals and/or patients. This study also introduced application software based on model C for the therapeutic drug monitoring of linezolid.

We study the voltage dependence of the SET time statistics of bipolar resistive switching in a tantalum oxide thin film. Weibull analysis reveals that the SET time statistics exhibit a crossover from deterministic to stochastic nature in a single cell as the amplitude of the applied voltage is lowered. While the Joule heating effect has a general contribution to the SET physics in both deterministic and stochastic cases, the magnitude of the positive feedback cycle of the Joule heating determines the statistical nature. Sufficient feedback effect under a voltage of large amplitude increases the SET probability with time, resulting in the SET time distribution with a deterministic nature. When amplitude of the applied voltage is small, on the other hand, the feedback effect is weak hence the SET process is controlled by randomness of the cell condition. In this case, the SET time is totally unpredictable hence its statistics has a stochastic nature. Since the crossover between the deterministic and stochastic regimes is found to occur based on the electric field rather than the current, we argue that it stems from a field-driven redox reaction at the tantalum oxide/tantalum interface.

Building compact and efficient reinforcement learning (RL) systems for mobile deployment requires departure from the von-Neumann computing architecture and embracing novel in-memory computing, and local learning paradigms. We exploit nano-scale ferroelectric tunnel junction (FTJ) memristors with inherent analogue stochastic switching arranged in selector-less crossbars to demonstrate an analogue in-memory RL system, which, via a hardware-friendly algorithm, is capable of learning behavior policies. We show that commonly undesirable stochastic conductance switching is actually, in moderation, a beneficial property which promotes policy finding via a process akin to random search. We experimentally demonstrate path-finding based on reinforcement, and solve a standard control problem of balancing a pole on a cart via simulation, outperforming similar deterministic RL systems.

Low power neural network hardware and its new applications have been explored to exploit its inherent advantage of artificial intelligence in comparison with humans. One such application, long-term change detection, is proposed and presented in this live demonstration. Owing to the low power operation in static analog/digital-mixed neural network circuits, our system using them can detect a change of human-friendly information, e.g., handwritten digits, whereas humans have difficulty noticing a gradual change over the long-term.

A low-power and stable “static-type” neural network (NN) circuit based on CMOS and resistive synaptic devices was developed and evaluated. The circuit is composed of a comparator as a firing function for binary output, current sources, cross switches for inputs, and variable resistors for synaptic weights. Nonlinearity analysis of operating points in such circuits was performed in terms of the types of current sources and the resistance-change ratios of the variable resistors. The operation window to realize both operation stability and low power consumption of less than 1 mW for 1024 synapses was thus clarified. To improve the NN performance, memristive synaptic devices with WOx/MgO were fabricated and characterized in terms of spike-timing-dependent plasticity and nonlinear switching for learning.

Linezolid (LZD) is an antimicrobial that is commonly used for treatment of vancomycin-resistant Enterococci and methicillin-resistant Staphylococcus aureus infections. However, the development of thrombocytopenia, one of the most frequent adverse side effects of this antimicrobial, can lead to discontinuation of LZD treatment. While clinical studies indicate that risk factors for the development of LZD-induced thrombocytopenia include treatment for >14 consecutive days, renal dysfunction, and chronic liver disease, the fundamental mechanism governing the pathogenesis of this disorder remains unclear. In this study, we aimed to elucidate the mechanism of LZD-induced thrombocytopenia by investigating the impact of LZD treatment on platelet destruction and production using rat platelet-rich plasma (PRP) and human immortalized cell lines, respectively. Compared to the control population, an increase in lactate dehydrogenase release was not detected upon exposure of rat PRP to varying concentrations of LZD, indicating that this compound is not cytotoxic towards platelets. Meanwhile, LZD treatment resulted in a significant dose-dependent increase in the proliferation of HEL human erythroleukemia and MEG-01 human megakaryoblast cells in vitro, but did not influence the differentiation of these cell lines. Lastly, LZD treatment yielded elevated levels of phosphorylation of myosin light chain 2 (MLC2), which regulates platelet release, in MEG-01 cells. Based on these results, we speculate that LZD induces thrombocytopenia by promoting MLC2 phosphorylation and thereby suppressing the release of platelets from mature megakaryocytes. These findings provide the first insight into the mechanism of LZD-mediated thrombocytopenia and may facilitate the development of strategies to treat and/or prevent this disease.

Artificial neural network (NN) circuits are proposed and analyzed from the viewpoints of flexibility and robustness based on memristive devices. The typical 3-layered fully-connected NN is a primitive unit of Deep Learning (DL) and plays a key role in determining DL performance. We propose three types of NN structure at a circuit level — fully-digital, analog/digital mixed, fully-analog — focusing on digital or analog calculations such as a multiplier accumulator, and examine their figures of merit. One key property of the analog/digital-mixed structure is demonstrated on reconfigurable analog ICs. Important aspects regarding neuromorphic computing are highlighted in our experiments on spike-timing-dependent plasticity on memristive devices and non-linear characteristics.

The development of quantum annealing machines (QAMs) based on superconducting qubits has progressed greatly in recent years and these machines are now widely used in both academia and commerce. On the other hand, QAMs based on semiconductor nanostructures such as quantum dots (QDs) appear to be still at the initial elementary research stage because of difficulty in controlling the interaction between qubits. In this paper, we review a QAM based on a semiconductor nanostructures such as floating gates (FGs) or QDs from the viewpoint of the integration of qubits. We theoretically propose the use of conventional high-density memories such as NAND flash memories for the QAM rather than the construction of a semiconductor qubit system from scratch. A large qubit system will be obtainable as a natural extension of the miniaturization of commercial-grade electronics, although further effort will likely be required to achieve high-quality qubits., Comment: 10 pages, 9 figures. arXiv admin note: text overlap with arXiv:1706.07565

The distribution of SET switching time of bipolar switching tantalum oxide thin films is studied using pulse measurement techniques. SET switching times are measured by repeating SET and RESET operations in a single cell. It is found that the distribution measured with a high RESET voltage can be well described by a steep Weibull distribution with a shape parameter >1, but lowering the RESET voltage results in a broad distribution at high cumulative frequencies. Statistical analysis shows that this broadening of the distribution can be attributed to the variation of initial conditions for SET, whereas a steep Weibull distribution points to an aging process leading to SET under a voltage stress. It is also shown that although the power of the leakage current before SET determines the fastest limit of the SET switching speed, those SET in the broadened distributions are delayed due to gradual current increase prior to the SET. The waiting time for the start of the gradual current increase has a correlation with the power of the leakage current, showing that Joule heating effect is still a significant factor in the SET mechanism even if the SET time distribution is broadened due to the variation of initial states programmed by low RESET voltages.

Single-electron effects have been widely investigated as a typical physical phenomenon in nanoelectronics. The single-electron effect caused by trap sites has been observed in many devices. In general, traps are randomly distributed and not controllable; therefore, different current--voltage characteristics are observed through traps even in silicon transistors having the same device parameters (e.g., gate length). This allows us to use single-electron effects as fingerprints of chips. In this study, we analyze the single-electron effect of traps in conventional silicon transistors and show the possibility of their use as fingerprints of chips through image recognition algorithms. Resonant tunneling parts in the Coulomb diagram can also be used to characterize each device. These results show that single-electron effects can provide a quantum version of a physically unclonable function (quantum-PUF)., Comment: 5 pages, 8 figures

Gate-first high-k/metal-gate n-MOSFETs with a deep subnanometer (sub-nm) equivalent oxide thickness (EOT) of 0.58 nm were successfully fabricated with Schottky source/drain contacts using a low-temperature process. The key to achieving such a thin EOT is the use of a low-temperature process for the NiSi Schottky source/drain formation. A sulfur implantation technique was used to overcome the Schottky barrier height between Si and NiSi, which is the main problem in NiSi Schottky source/drain fabrication. The advantage of Schottky source/drain MOSFETs over conventional source/drain MOSFETs was successfully demonstrated.

In this letter, we discuss the kinetics of the resistive switch effect in tantalum oxide thin films. The time to switch from the high resistance state to the low resistance state (SET) was measured by using pulse measurement technique. It was found that the SET switching time has a clear relationship with the power of the leakage current flowing under the applied voltage before the SET. Although the dependence of the SET time on the applied voltage differ from cell to cell, this relationship is universal among the different cells. This implies that the Joule heating effect is the dominant driving factor of the SET, rather than the applied voltage. In addition, this relationship can account for the faster switching speed at an elevated temperature.