Boosting-Crystal Graph Convolutional Neural Network for Predicting Highly Imbalanced Data: A Case Study for Metal-Insulator Transition Materials.

Applying machine-learning techniques for imbalanced data sets presents a significant challenge in materials science since the underrepresented characteristics of minority classes are often buried by the abundance of unrelated characteristics in majority of classes. Existing approaches to address this focus on balancing the counts of each class using oversampling or synthetic data generation techniques. However, these methods can lead to loss of valuable information or overfitting. Here, we introduce a deep learning framework to predict minority-class materials, specifically within the realm of metal-insulator transition (MIT) materials. The proposed approach, termed boosting-CGCNN, combines the crystal graph convolutional neural network (CGCNN) model with a gradient-boosting algorithm. The model effectively handled extreme class imbalances in MIT material data by sequentially building a deeper neural network. The comparative evaluations demonstrated the superior performance of the proposed model compared to other approaches. Our approach is a promising solution for handling imbalanced data sets in materials science.

First-Principles Investigations on the Semiconductor-to-Metal Phase Transition of 2D Si2Te3 for Reversible Resistive Switching.

Si2Te3 is attracting attention due to its compatibility with Si technology while still showing advantages as a two-dimensional layered material. Although recent experimental studies have observed the resistive switching process in Si2Te3-based memristors, the mechanism has not been clearly identified. In this study, first-principles density functional theory calculations are employed to understand the relationship between the phase transition of Si2Te3 and the reversible resistive switching of the Si2Te3-based memristor. Our calculation results show that although semiconducting Si2Te3 is energetically more stable than two metallic Si2Te3 phases (α and ß), two metallic Si2Te3 can be energetically stabilized by excess holes. The enhanced energetic preference of two metallic Si2Te3 by excess holes is explained by the reduced occupation of antibonding states between Si and Te. Our study finds that the energy barrier for the phase transition between semiconducting Si2Te3 and α-metallic Si2Te3 varies significantly by excess charge carriers so the phase transition can be directly connected to the reversible resistive switching of the Si2Te3-based memristor under external bias. Our finding will serve as a cornerstone for optimizing the resistive switching process of the Si2Te3-based memristor.

Semantic Segmentation Using Pixel-Wise Adaptive Label Smoothing via Self-Knowledge Distillation for Limited Labeling Data.

To achieve high performance, most deep convolutional neural networks (DCNNs) require a significant amount of training data with ground truth labels. However, creating ground-truth labels for semantic segmentation requires more time, human effort, and cost compared with other tasks such as classification and object detection, because the ground-truth label of every pixel in an image is required. Hence, it is practically demanding to train DCNNs using a limited amount of training data for semantic segmentation. Generally, training DCNNs using a limited amount of data is problematic as it easily results in a decrease in the accuracy of the networks because of overfitting to the training data. Here, we propose a new regularization method called pixel-wise adaptive label smoothing (PALS) via self-knowledge distillation to stably train semantic segmentation networks in a practical situation, in which only a limited amount of training data is available. To mitigate the problem caused by limited training data, our method fully utilizes the internal statistics of pixels within an input image. Consequently, the proposed method generates a pixel-wise aggregated probability distribution using a similarity matrix that encodes the affinities between all pairs of pixels. To further increase the accuracy, we add one-hot encoded distributions with ground-truth labels to these aggregated distributions, and obtain our final soft labels. We demonstrate the effectiveness of our method for the Cityscapes dataset and the Pascal VOC2012 dataset using limited amounts of training data, such as 10%, 30%, 50%, and 100%. Based on various quantitative and qualitative comparisons, our method demonstrates more accurate results compared with previous methods. Specifically, for the Cityscapes test set, our method achieved mIoU improvements of 0.076%, 1.848%, 1.137%, and 1.063% for 10%, 30%, 50%, and 100% training data, respectively, compared with the method of the cross-entropy loss using one-hot encoding with ground truth labels.

Surface Diffusion and Epitaxial Self-Planarization for Wafer-Scale Single-Grain Metal Chalcogenide Thin Films.

Although wafer-scale single-grain thin films of 2D metal chalcogenides (MCs) have been extensively sought after during the last decade, the grain size of the MC thin films is still limited in the sub-millimeter scale. A general strategy of synthesizing wafer-scale single-grain MC thin films by using commercial wafers (Si, Ge, GaAs) both as metal source and epitaxial collimator is presented. A new mechanism of single-grain thin-film formation, surface diffusion, and epitaxial self-planarization is proposed, where chalcogen elements migrate preferentially along substrate surface and the epitaxial crystal domains flow to form an atomically smooth thin film. Through synchrotron X-ray diffraction and high-resolution scanning transmission electron microscopy, the formation of single-grain Si2 Te3 , GeTe, GeSe, and GaTe thin films on (111) Si, Ge, and (100) GaAs is verified. The Si2 Te3 thin film is used to achieve transfer-free fabrication of a high-performance bipolar memristive electrical-switching device.

Development of a naftopidil-chitosan-based fumaric acid solid dispersion to improve the dissolution rate and stability of naftopidil.

Naftopidil (NAF), an α1-adrenoceptor antagonist, is administered as a treatment for benign prostatic hyperplasia; however, according to the Biopharmaceutical Classification System (BCS IV), it is a poorly-soluble drug that exhibits poor permeability. We aimed to increase the dissolution (%) of NAF by adding chitosan to a polymer-free formulation. Compared to the formulation prepared using Flivas®, at 60 min, the solid dispersion (SD) formulation containing NAF, fumaric acid, chitosan, and US2® in a 1:1:2:1 weight ratio improved the dissolution (%) of NAF in distilled water, pH 1.2 media, pH 4.0 and pH 6.8 buffers by 27.2-, 1.2-, 1.1- and 6.5-fold, respectively. The physicochemical properties of the SD1 formulation were also found to be altered, including its thermal properties, crystal intensity, and chemical interaction. As a result, the hydrogen bonding that occurs between NAF and fumaric acid was identified as a major factor in the increase in NAF dissolution (%). Further, chitosan was observed to contribute to the stability of NAF and SD1, which was assessed over a 3-month period. To our knowledge, this is the first study to employ a polymer-free system to improve the solubilization of NAF.

How good is cola for dissolution of gastric phytobezoars?

AIM: To evaluate the efficacy of cola treatment for gastric phytobezoars, including diospyrobezoars. METHODS: A total of 17 patients (range: 48 to 78 years) with symptomatic gastric phytobezoars treated with cola and adjuvant endoscopic therapy were reviewed. Three liters of cola lavage (10 cases) or drink (7 cases) were initially used, and then endoscopic fragmentation was done for the remnant bezoars by using a lithotripsy basket or a polypectomy snare. The overall success of dissolving a gastric phytobezoars with using three liters of cola and the clinical and endoscopic findings were compared retrospectively between four cases of complete dissolution by using only cola and 13 cases of partial dissolution with cola. RESULTS: After 3 L of cola lavage or drinking, a complete dissolution of bezoars was achieved in four patients (23.5%), while 13 cases (76.5%) were only partially dissolved. Phytobezoars (4 of 6 cases) were observed more frequently than diospyrobezoars (0 of 11) in the group that underwent complete dissolution (P = 0.006). Gender, symptom duration, size of bezoar and method of cola administration were not significantly different between the two groups. Twelve of 13 patients with residual bezoars were completely treated with a combination of cola and endoscopic fragmentation. CONCLUSION: The rate of complete dissolution with three liters of cola was 23.5%, but no case of diospyrobezoar was completely dissolved using this method. However, pretreatment with cola may be helpful and facilitate endoscopic fragmentation of gastric phytobezoars.

Pancreatic tuberculosis masquerading as pancreatic serous cystadenoma.

Solitary pancreatic involvement of tuberculosis is rare, especially in an immunocompetent individual, and it may be misdiagnosed as pancreatic cystic neoplasms. Pancreatic cystic neoplasms are being identified in increasing numbers, probably because of the frequent use of radiology and advances in endoscopic techniques. However, they are composed of a variety of neoplasms with a wide range of malignant potential, and it is often difficult to differentiate pancreatic tuberculosis mimicking cystic neoplasms from benign or malignant pancreatic cystic neoplasms. Non-surgical diagnosis of pancreatic tuberculosis is inconclusive and continues to be a challenge in many cases. If so, then laparotomy should be employed to establish the diagnosis. Therefore, pancreatic tuberculosis should be kept in mind during the differential diagnosis of solitary cystic masses in the pancreas. We report a patient who had solitary pancreatic tuberculosis masquerading as pancreatic serous cystadenoma.

Antiviral efficacy of adefovir dipivoxil versus lamivudine in patients with chronic hepatitis B sequentially treated with lamivudine and adefovir due to lamivudine resistance.

AIM: To compare the antiviral efficacy of adefovir (ADV) in lamivudine (LMV)-resistant patients with LMV treatment in nucleoside-naive patients, using serum samples collected sequentially during the course of treatment progressing from LMV to ADV. METHODS: Forty-four patients with chronic hepatitis B (CHB) were included. The patients were initially treated with LMV and then switched to ADV when LMV resistance developed. Antiviral efficacy was assessed by measuring the following: reduction in serum HBV DNA from baseline, HBV DNA negative conversion (defined as HBV DNA being undectable by the hybridization assay), and HBV DNA response (either HBV DNA level or= 2 log10 reduction from baseline HBV DNA level). RESULTS: After two and six months of treatment, HBV DNA reduction was greater with LMV compared to ADV treatment (P = 0.021). HBV DNA negative conversion rates were 64% and 27% after one month of LMV and ADV treatment respectively (P = 0.001). Similarly, HBV DNA response rates were 74% and 51% after two months of LMV and ADV treatment respectively (P = 0.026). The time taken to HBV DNA negative conversion and to HBV DNA response were both delayed in ADV treatment compared with LMV. CONCLUSION: The antiviral efficacy of ADV in LMV-resistant patients is slower and less potent than that with LMV in nucleoside-naive patients during the early course of treatment.

Removal of press-through-packs impacted in the upper esophagus using an overtube.

Foreign bodies in the upper esophagus should be removed as soon as possible to avoid serious complications. However, removals of foreign bodies in the upper esophagus are very difficult, especially if they have sharp edges, such as press-through-packs (PTPs). We experienced four cases of the impacted PTPs in the upper esophagus which was successfully extracted endoscopically with the overtube. Because two edges of PTPs were so firmly impacted in the esophageal wall in all cases, the PTPs were not movable in the upper esophagus. However, after insertion of the overtube, PTPs became movable and were successfully extracted and no serious complications occurred after extraction of PTPs. In one case, insertion of the overtube rapidly expanded the upper esophagus and PTP progressed to the gastric cavity and it could be extracted with the endoscopic protector hood. The endoscopic removal with the overtube was a simple, safe and effective technique for the removal of the impacted PTPs in upper esophagus.