Continual learning with attentive recurrent neural networks for temporal data classification.

Continual learning is an emerging research branch of deep learning, which aims to learn a model for a series of tasks continually without forgetting knowledge obtained from previous tasks. Despite receiving a lot of attention in the research community, temporal-based continual learning techniques are still underutilized. In this paper, we address the problem of temporal-based continual learning by allowing a model to continuously learn on temporal data. To solve the catastrophic forgetting problem of learning temporal data in task incremental scenarios, in this research, we propose a novel method based on attentive recurrent neural networks, called Temporal Teacher Distillation (TTD). TTD solves the catastrophic forgetting problem in an attentive recurrent neural network based on three hypotheses, namely Rotation Hypothesis, Redundant Hypothesis, and Recover Hypothesis. Rotation Hypothesis and Redundant hypotheses could cause the attention shift phenomenon, which degrades the model performance on the learned tasks. Moreover, not considering the Recover Hypothesis increases extra memory usage in continuously training different tasks. Therefore, the proposed TTD based on the above hypotheses complements the inadequacy of the existing methods for temporal-based continual learning. For evaluating the performance of our proposed method in task incremental setting, we use a public dataset, WIreless Sensor Data Mining (WISDM), and a synthetic dataset, Split-QuickDraw-100. According to experimental results, the proposed TTD significantly outperforms state-of-the-art methods by up to 14.6% and 45.1% in terms of accuracy and forgetting measures, respectively. To the best of our knowledge, this is the first work that studies continual learning in real-world incremental categories for temporal data classification with attentive recurrent neural networks and provides the proper application-oriented scenario.

Evidence of the need for modified well-stirred model in vitro to in vivo extrapolation.

In vitro to in vivo extrapolation (IVIVE), an approach for hepatic clearance (CLH) prediction used worldwide, remains controversial due to systematic underprediction. Among the various probable factors, the original assumption of the hepatic mathematical model (i.e., the well-stirred model, WSM) may become problematic, leading to the underestimation of drug CLH. Having a similar prerequisite that the well-stirred conditions are homogenous with perfectly mixed reactants, but using a different driving concentration, the modified well-stirred model (MWSM) stands apart from the WSM. However, we believe that both models should coexist so that the entire well-stirred scenario can be completely illustrated. Consequently, we collected published data from the literature and employed a logistic regression method to differentiate the optimal timing of use between WSM and MWSM in drug CLH prediction. Generally, variances adopted in the regression, including partition coefficient (logP), fraction unbound (fu), volumes of distribution at steady-state (Vss), and mean residence time (MRT), corresponded to our assumption when protein-facilitated uptake was considered. Furthermore, a new empirical approach was introduced to allow practical use of the MWSM. The results showed that this model could provide a more precise prediction compared to previous empirical approaches. Therefore, these preliminary results not only delineated a more detailed structure and mechanism of MWSM but also highlighted its necessity and potential.

Precisely adjusting the hepatic clearance of highly extracted drugs using the modified well-stirred model.

Hepatic clearance has been widely studied for over 50 yr. Many models have been developed using either theoretical or empirical tests to predict drug metabolism. The well-stirred, parallel-tube, and dispersion metabolic models have been extensively discussed. However, to our knowledge, these models cannot fully describe all relevant scenarios in hepatic clearance. We addressed this issue using the isolated perfused rat liver technique with minor modifications. Diazepam was selected to illustrate different levels of drug plasma-protein binding by changing the added concentration of human serum albumin. The free fractions of diazepam at different albumin concentrations were assayed by rapid equilibrium dialysis. The experimental data provide new insights concerning an accepted formula used to describe hepatic clearance. Regarding drug concentrations passing through the liver, the driving force concentration (CH,ss) in terms of Cin (influx in the liver) or Cout (efflux from the liver) needs to be carefully considered when determining drug hepatic and intrinsic clearances. The newly established model, termed the modified well-stirred model, which was derived from the original formula, successfully estimated hepatic drug metabolism. Using the modified well-stirred model, a theoretical driving force concentration of diazepam passing through the liver was evaluated. The model was further used to assess the predictability of in vitro to in vivo extrapolation. This study was not intended to refute the existing models, but rather to augment them using experimental data. The results stress the importance of proper calculation of dose when the drug clearance deviates from the prediction of the well-stirred model.

Three-layer heterogeneous mammographic phantoms for Monte Carlo simulation of normalized glandular dose coefficients in mammography.

Normalized glandular dose (DgN) coefficients obtained using homogeneous breast phantoms are commonly used in breast dosimetry for mammography. However, glandular tissue is heterogeneously distributed in the breast. This study aimed to construct three-layer heterogeneous mammographic phantoms (THEPs) to examine the effect of glandular distribution on DgN coefficient. Each layer of THEPs was set to 25%, 50%, or 75% glandular fraction to emulate heterogeneous glandular distribution. Monte Carlo simulation was performed to attain mean glandular dose (MGD) and air kerma at 22-36 kVp and W/Al, W/Rh, and W/Ag target-filter combinations. The heterogeneous DgN coefficient was calculated as functions of the mean glandular fraction (MGF), breast thickness, tube voltage, and half-value layer. At 50% MGF, the heterogeneous DgN coefficients for W/Al, W/Rh, and W/Ag differed by 40.3%, 36.7%, and 31.2%. At 9-cm breast thickness, the DgN values of superior and inferior glandular distributions were 25.4% higher and 29.2% lower than those of uniform distribution. The proposed THEPs can be integrated with conventional breast dosimetry to consider the heterogeneous glandular distribution in clinical practice.

Right Ventricle Mimics Right Atrium at First Glance: A Rare Case of Congenital Right Sided Partial Pericardial Defect.

Among heart irregularities, congenital pericardial defect is an unusual anomaly, and is typically left dominant. However, cases of right pericardial defect with heart herniation are extremely rare. This is a case of congenital right pericardial defect with herniation of the right ventricle free wall and right ventricular outflow tract. The patient is asymptomatic and refused further intervention but even indolent discomfort underscores the risks of iatrogenic injuries to the heart and sudden death caused by mechanical pathogenesis due to changes in anatomical positions of the cardiac structures.

Seasonal variation of Legionella in Taiwan's reservoir and its relationships with environmental factors.

In this study, the presence of Legionella in major water reservoirs of Taiwan was examined with respect to seasonal variation, geographical variation, and water quality parameters using TaqMan real-time qPCR. Water samples were collected quarterly at 19 reservoirs in Taiwan between November 2012 and August 2013. The detection rate for Legionella was 35.5% (27/76), and Legionella was detected in all seasons. The Legionella concentration was relatively high in spring and summer, reaching 3.86 × 10(8) and 7.35 × 10(8) cells/L, respectively. By sampling the area, Legionella was detected at a higher proportion in reservoirs in the northern and southern areas, and the difference was consistent in all seasons. Significant association was found between detection of Legionella and various water quality parameters, including conductivity, chlorophyll a, and dissolved oxygen (Mann-Whitney U test, P < 0.05). Results of Spearman rank test showed negative correlation for Legionella detection with pH (P = 0.030, R = -0.497) and dissolved oxygen (P = 0.007, R = -0.596) in fall and positive correlation with Carlson's trophic state index (P = 0.049, R = 0.457) in spring. The identified species included Legionella pneumophila and Legionella drancourtii. The detection of Legionella in reservoirs was indicative of a potential public health risk and should be further evaluated.

Prevalence, quantification, and typing of human adenoviruses detected in river water in Taiwan.

The prevalence of human adenoviruses (HAdV) in river waters was investigated in this study. Water samples were collected from 13 rivers in Taiwan, concentrated, and assessed for the presence of HAdVs using nested polymerase chain reaction (PCR). Human AdV positive samples were then subjected to real-time PCR (qPCR) to quantify the viral genomes and further subjected to primer-based genotyping to identify the various serotypes present. For each water sample, several water quality parameters were evaluated, including heterotrophic plate count, total coliform, Escherichia coli, water temperature, pH, conductivity, and dissolved oxygen. Among the 13 rivers examined, four rivers (30.8 %) were found to contain HAdVs. The major genotype was F species HAdV serotype 41. The mean HAdVs concentrations ranged from 6.10 × 10(2) to 8.51 × 10(2) copies/L. No significant differences were observed between the presence of HAdVs, and all of the water quality parameters evaluated (heterotrophic plate count, total coliform, E. coli, water temperature, pH, conductivity, and dissolved oxygen). Given the potential health risks posed by the presence of enteric viruses in environmental waters, further assessment is desirable with respect to possible sources, virus transport, and survival of viruses in the aquatic environment.

Evaluation of immunomagnetic separation for the detection of Salmonella in surface waters by polymerase chain reaction.

Salmonella spp. is associated with fecal pollution and capable of surviving for long periods in aquatic environments. Instead of the traditional, time-consuming biochemical detection, polymerase chain reaction (PCR) allows rapid identification of Salmonella directly concentrated from water samples. However, prevalence of Salmonella may be underestimated because of the vulnerability of PCR to various environmental chemicals like humic acid, compounded by the fact that various DNA polymerases have different susceptibility to humic acid. Because immunomagnetic separation (IMS) theoretically could isolate Salmonella from other microbes and facilitate removal of aquatic PCR inhibitors of different sizes, this study aims to compare the efficiency of conventional PCR combined with immunomagnetic separation (IMS) for Salmonella detection within a moderately polluted watershed. In our study, the positive rate was increased from 17.6% to 47% with nearly ten-fold improvement in the detection limit. These results suggest the sensitivity of Salmonella detection could be enhanced by IMS, particularly in low quality surface waters. Due to its effects on clearance of aquatic pollutants, IMS may be suitable for most DNA polymerases for Salmonella detection.

Surveillance and evaluation of the infection risk of free-living amoebae and Legionella in different aquatic environments.

Free-living amoebae (FLA) are ubiquitous in various aquatic environments. Several amoebae species are pathogenic and host other pathogens such as Legionella, but the presence of FLA and its parasites as well as the related infection risk are not well known. In this study, the presence of pathogenic FLA and Legionella in various water bodies was investigated. Water samples were collected from a river, intake areas of drinking water treatment plants, and recreational hot spring complexes in central and southern Taiwan. A total of 140 water samples were tested for the presence of Acanthamoeba spp., Naegleria spp., Vermamoeba vermiformis, and Legionella. In addition, phylogenetic characteristics and water quality parameters were also assessed. The pathogenic genotypes of FLA included Acanthamoeba T4 and Naegleria australiensis, and both were abundant in the hot spring water. In contrast, Legionella pneumophila was detected in different aquatic environments. Among the FLA assessed, V. vermiformis was most likely to coexist with Legionella spp. The total bacteria level was associated with the presence of FLA and Legionella especially in hot spring water. Taken together, FLA contamination in recreational hot springs and drinking water source warrants more attention on potential legionellosis and amoebae infections.

Bilateral ultrasound-guided continuous ilioinguinal-iliohypogastric block for pain relief after cesarean delivery.

We present three cases in which continuous ilioinguinal-iliohypogastric nerve block with 0.2% ropivacaine, together with oral ibuprofen, was used to provide analgesia after cesarean delivery. The catheters were placed under ultrasound guidance in the plane between the internal oblique and transversus abdominis muscles on both sides of the abdomen. Numeric pain rating was used for the assessment of postoperative pain. Low pain scores, minimal use of supplemental opioid, and the absence of nausea and vomiting suggests that continuous ilioinguinal-iliohypogastric nerve blockade deserves further study as a possible component of multimodal analgesia after cesarean delivery.