Search

In the field of natural language processing, artificial intelligence (AI) technology has been utilized to solve various problems, such as text classification, similarity measurement, chatbots, machine translation, and machine reading comprehension. Significant advancements have been made in complex and rule-intensive natural language processing through deep learning, in which machines directly learn patterns. Machine reading comprehension, a natural language processing task, involves machines understanding questions and paragraphs to find answers within paragraphs. In 2019, bidirectional encoder representations from transformers (BERT) and a robust optimized BERT pretraining approach (RoBERTa) were introduced. They were then optimized for pretraining and fine-tuning, resulting in significant advancements. RoBERTa outperformed BERT in terms of training speed and performance by increasing the pretraining data and batch sizes, employing dynamic masking, and eliminating the next sentence prediction task. In RoBERTa, machine reading comprehension involves simultaneously inputting questions and paragraphs. However, this simultaneous input method suffers from the attention separate representation (ASP) problem, in which the attention distribution between the question and the paragraph spreads widely across keywords. This study proposed two methods to address the ASP problem. The existing input format, question-paragraph, was changed to three independent inputs: question, paragraph, and concatenated RoBERTa outputs. The concatenated matrix was then transformed into two matrices, and a machine reading comprehension algorithm using co-attention was proposed. An ablation study was conducted to evaluate and analyze the model’s performance, comprehension, and design efficiency. According to the experimental results, the proposed method improved the EM by 0.9% and F1 by 1.0% compared to the existing methods. Consequently, the learning performance was enhanced through attention concentration and co-attention, and it could demonstrate much better performance compared to the existing models.

Vision Transformer (ViT) has been proposed as a new image recognition method in the field of computer vision. ViT applies a Transformer structure with excellent performance in the field of natural language processing to recognize images. Unlike existing Convolutional Neural Network (CNN) models, ViT can achieve State-Of-The-Art (SOTA) image recognition without inputting Inductive Biases into the model, demonstrating that the Transformer is a useful structure in the field of computer vision. However, ViT requires large datasets such as ImageNet-21K and Joint Foto Tree (JFT) for learning. In addition, it takes a lot of time to train. Moreover, there is a problem that location information is lost by inputting images in patch units. To improve such issues, many models are being proposed. In this paper, a new model is proposed by restructuring the Convolutional neural networks Meet vision Transformers (CMT) model by applying Coordinate Attention Block, a CNN model, to improve problems of the Vision Transformer family of models. The proposed model combines Transformer, which has shown excellent performance in Long Range, and CNN, which has shown excellent performance in Local Feature, to achieve higher performance than existing models. We also compared performance of the proposed model with those of existing models with relatively small datasets such as Canadian Institute For Advanced Research-10 (CIFAR-10), Self-Taught Learning-10 (STL-10), and Tiny-ImageNet to facilitate researchers’ access to the evaluation. Despite being restructured from the smallest CMT-Tiny model, the proposed model showed better accuracy than CMT-Tiny, CMT-XS, CMT-S, and CMT-B models with CIFAR-10, STL-10, and Tiny-ImageNet datasets. The proposed model showed an accuracy of 90.21% with the CIFAR-10 dataset, higher than existing CMT models except for the CMT-S model with an accuracy of 90.6%. It had the lowest loss value of 0.3967. The proposed model is expected to be utilized as a backbone in Object Detection and Segmentation fields in the future.

The base deficit (B), international normalized ratio (I), and Glasgow coma scale (GCS) (BIG) score is useful in predicting mortality in pediatric trauma patients; however, studies on the use of BIG score in adult patients with trauma are sparse. In addition, studies on the correlation between the BIG score and massive transfusion (MT) have not yet been conducted. This study aimed to evaluate the predictive value of BIG score for mortality and the need for MT in adult trauma patients. This retrospective study used data collected between 2016 and 2020 at our hospital’s trauma center and registry. The predictive value of BIG score was compared with that of the Injury Severity Score (ISS) and Revised Trauma Score (RTS). Logistic regression analysis was carried out to assess whether BIG score was an independent risk factor. Receiver operating characteristic (ROC) curve analysis was performed, and predictive values were evaluated by measuring the area under the ROC curve (AUROC). In total, 5,605 patients were included in this study. In logistic regression analysis, BIG score was independently associated with in-hospital mortality (odds ratio (OR): 1.1859; 95% confidence interval (CI): 1.1636–1.2086) and MT (OR: 1.0802; 95% CI: 1.0609–1.0999). The AUROCs of BIG score for in-hospital mortality and MT were 0.852 (0.842–0.861) and 0.848 (0.838–0.857), respectively. Contrastingly, the AUROCs of ISS and RTS for in-hospital mortality were 0.795 (0.784–0.805) and 0.859 (0.850–0.868), respectively. Moreover, AUROCs of ISS and RTS for MT were 0.812 (0.802–0.822) and 0.838 (0.828–0.848), respectively. The predictive value of BIG score for mortality and MT was significantly higher than that of the ISS. The BIG score also showed a better AUROC for predicting in-hospital mortality compared with RTS. In conclusion, the BIG score is a useful indicator for predicting mortality and the need for MT in adult trauma patients.

Background/Aims: Although localized lymphoid hyperplasia (LLH) of the rectum is occasionally observed, its clinical implications are unclear. This study aimed to investigate the clinical course and significance of LLH of the rectum. Methods: We identified 65 patients diagnosed with LLH of the rectum using a histopathologic examination and who received follow-up endoscopies between January 2009 and June 2015. Patients with a history of inflammatory bowel disease, lymphoma, familial adenomatous polyposis, or uncontrolled malignancy and patients who underwent scar biopsy after endoscopic resection or surgery were excluded. Endoscopic findings and clinical courses were analyzed. Results: During the median follow-up of 31 months (interquartile range, 19 to 40 months), 81.5% (53/65) of LLHs of the rectum were resolved. Clinically significant diseases, including ulcerative colitis (UC, n=5) and mucosa-associated lymphoid tissue (MALT) lymphoma (n=1), were diagnosed in 9.2% of patients (6/65). The other six patients showed no significant changes in the lesion (n=3) or a waxing and waning appearance (n=3). According to endoscopic findings, all of the 47 polypoid types showed resolution or waxing and waning patterns. Five of the 11 nodular types (45.5%) developed into UC. One of the seven submucosal tumor (SMT)-like types (14.3%) developed into MALT lymphoma. Conclusions: LLH of the rectum with persistent symptoms or the endoscopic appearance of the nodular or SMT-like type may lead to clinically significant disease. Risk stratification according to endoscopic findings and careful surveillance are required for these lesions.

Abstract Background In plants, the key roles played by RopGEF-mediated ROP signaling in diverse processes, including polar tip growth, have been identified. Despite their important roles in reproduction, a comprehensive analysis of RopGEF members has not yet been performed in rice (Oryza sativa). To determine whether RopGEF regulators are involved in rice pollen tube growth, we performed genome-wide analysis of this family in rice. Results Phylogenomic and meta-expression analysis of eleven RopGEFs in rice showed that four genes were preferentially expressed in mature pollen. These four genes contain the plant-specific Rop nucleotide exchanger (PRONE) domain and possible phosphorylated residues, suggesting a conserved role in polar tip growth with Arabidopsis thaliana. In subcellular localization analysis of the four RopGEFs through tobacco (Nicotiana benthamiana) infiltration, four proteins were predominantly identified in plasma membrane. Moreover, double mutants of RopGEF2/8 exhibited reduced pollen germination, causing partial male sterility. These genes possess unique cis-acting elements in their promoters compared with the other RopGEF genes. Conclusions In this study, four RopGEF genes were identified as pollen-specific gene in eleven members of rice, and the expression pattern, promoter analysis, and evolutionary relationship of the RopGEF family were studied compared with Arabidopsis. Our study indicated that four RopGEF genes might function during pollen germination in distinct subcellular localization. Our study could provide valuable information on the functional study of RopGEF in rice.

Background. Rotational thrombelastometry (ROTEM) has been used to evaluate the coagulation state, predict transfusion, and optimize hemostatic management in trauma patients. However, there were limited studies on whether the prediction value could be improved by adding the ROTEM parameter to the prediction model for in-hospital mortality and massive transfusion (MT) in trauma patients. Objective. This study assessed whether ROTEM data could improve the MT prediction model. Method. This was a single-center, retrospective study. Patients who presented to the trauma center and underwent ROTEM between 2016 and 2020 were included. The primary and secondary outcomes were massive transfusions and in-hospital mortality, respectively. We constructed two models using multivariate logistic regression with backward conditional stepwise elimination (Model 1: without the ROTEM parameter and Model 2: with the ROTEM parameter). The area under the receiver operating characteristic curve (AUROC) was calculated to assess the predictive ability of the models. Result. In total, 969 patients were included; 196 (20.2%) received MT. The in-hospital mortality rate was 14.1%. For MT, the AUROC was 0.854 (95% confidence interval [CI], 0.825–0.883) and 0.860 (95% CI, 0.832–0.888) for Model 1 and 2, respectively. For in-hospital mortality, the AUROC was 0.886 (95% CI, 0.857–0.915) and 0.889 (95% CI, 0.861–0.918) for models 1 and 2, respectively. The AUROC values for models 1 and 2 were not statistically different for either MT or in-hospital mortality. Conclusion. We found that the addition of the ROTEM parameter did not significantly improve the predictive power of MT and in-hospital mortality in trauma patients.

PURPOSE:We aimed to examine the utility of the Poison Severity Score (PSS) and Sequential Organ Failure Assessment (SOFA) score as early prognostic predictors of short-term outcomes in patients with carbon monoxide (CO) poisoning. We hypothesized that both the PSS and the SOFA score would be useful prognostic tools. METHODS:This was retrospective observational study of patients with CO poisoning who presented to the emergency department and were admitted for more than 24 hours. We calculated PSS, the initial SOFA score, a second (2nd) SOFA score, and a 24-hour delta SOFA score. The primary outcome was reported as the cerebral performance category (CPC) scale score at discharge. We classified those with CPC 1-2 as the good outcome group and those with CPC 3-5 as the poor outcome group. RESULTS:This study included 192 patients: 174 (90.6%) belonged to the good outcome group, whereas 18 (9.4%) belonged to the poor outcome group. The PSS (1.00 [0.00, 1.00] vs 3.00 [3.00, 3.00], p < 0.001), initial SOFA (1.00 [0.00, 2.00] vs 4.00 [3.25, 6.00], p < 0.001), 2nd SOFA score (0.00 [0.00, 1.00] vs 4.00 [3.00, 7.00], p < 0.001), and 24-hour delta SOFA score (-1.00 [-1.00, 0.00] vs 0.00 [-1.00, 1.00], p = 0.047) of the good outcome group were significantly higher than those of the poor outcome group. The areas under the receiver operating characteristic curve for PSS and the initial SOFA and 2nd SOFA scores were 0.977 (95% confidence interval [CI] 0.944-0.993), 0.945 (95% CI 0.903-0.973), and 0.978 (95% CI 0.947-0.994), respectively. CONCLUSION:The PSS, initial SOFA score, and 2nd SOFA score predict acute poor outcome accurately in patients with CO poisoning.

Background and Objectives: We examined the association between the baseline perfusion index (PI) and changes in intraoperative body temperature during general anesthesia. The PI reflects the peripheral perfusion state. The PI may be associated with changes in body temperature during general anesthesia because the degree of redistribution of body heat from the central to the peripheral compartment varies depending on the peripheral perfusion state. Materials and Methods: Thirty-eight patients who underwent brain surgery were enrolled in this study. The baseline PI and body temperature of the patients were measured on entering the operating room. Body temperature was recorded every 15 min after induction of anesthesia using an esophageal temperature probe. Univariate and multivariate logistic regression analyses were performed to identify the risk factors for intraoperative hypothermia. Results: Eighteen patients (47 %) developed hypothermia intraoperatively. The baseline PI was significantly lower among patients in the hypothermia group (1.8 ± 0.7) than among those in the normothermia group (3.0 ± 1.2) (P < 0.001). The baseline PI and body temperature were independently associated with intraoperative hypothermia (PI: odds ratio [OR], 0.270; 95% confidence interval [CI], 0.105–0.697; P = 0.007, baseline body temperature: OR, 0.061; 95% CI, 0.005–0.743; P = 0.028). Conclusions: This study showed that low baseline PI was the factor most related to the development of intraoperative hypothermia. Future studies should consider the PI as a predictor of intraoperative hypothermia.

Predisposing factors for venous thrombosis can be identified in the majority of patients with established venous thromboembolism (VTE). However, an obvious precipitant may not be identified during the initial evaluation of such patients. In the present case, a 47-year-old female presented to the emergency department of our hospital after ingesting multiple drugs. She had no VTE-related risk factors or previous episodes, nor any family history of VTE. After admission to the intensive care unit sudden hypoxemia developed, and during the evaluation cerebral, renal, and splenic infarctions with pulmonary embolisms were diagnosed. However, the sources of the emboli could not be identified by transthoracic echocardiography or computed tomography angiography. Protein C deficiency was identified several days later. We recommend that hypercoagulable states be taken into consideration, especially when unexplained thromboembolic events develop in multiple or unusual venous sites.

Colchicine poisoning is rare but can cause potentially life-threatening toxic complications such as hypovolemic shock, cardiovascular collapse and multiple organ failure. In this case report, we describe a case of a 20-year-old female who presented to the emergency department after suicidal ingestion of a toxic dose of colchicine. She developed thrombocytopenia, neutropenia and acute respiratory distress syndrome that required blood transfusion and administration of granulocyte colony stimulating factor for the prevention of infectious complications. With regard to the clinical manifestations of colchicine toxicity, we discussed suggested mechanisms.

BackgroundThe aim of this study was to determine whether the end-tidal concentration of desflurane would be affected by a breathing circuit system filter attached at two different positions in anesthetic breathing circuit systems.MethodsAn artificial lung was ventilated under five different conditions. The first group was without any filter or desflurane (n = 5, sham), the second was with desflurane but without any filter (n = 10, control), the third group had a bacterial filter on the expiratory limb (n = 10), and the fourth and fifth groups had a viral/bacterial filter added on the expiratory limb (n = 10) or at the Y-piece of the breathing circuit (n = 10), respectively. In all groups except the sham, administration of 10% desflurane was performed for 5 minutes and then stopped for 5 minutes.ResultsThe mean (SD) end-tidal concentration of desflurane for the groups described above peaked at 0 (0), 9.8 (0.1), 9.8 (0.1), 8.5 (0.1), and 6.7% (0.1) (P < 0.001), respectively. There was no difference in the desflurane concentrations and the expired tidal volume over time between the control and bacterial group, but there was a significant difference between the control and the fourth and fifth groups (P < 0.001).ConclusionsFilters can affect the expiratory desflurane concentration during anesthesia.

BackgroundBispectral index (BIS) monitoring reduces the cases of intraoperative awareness. Several factors can alter BIS readings without affecting the depth of anesthesia. We conducted a study to assess the impact of beach chair position (sitting position) on BIS readings.MethodsGeneral anesthesia was administered to 30 patients undergoing arthroscopic shoulder surgery. Patients were kept in neutral position (supine) for 10 minutes and BIS readings, mean arterial blood pressure, heart rate, end-tidal carbon dioxide, and end-tidal sevoflurane were recorded. Patients were then shifted to beach chair position. After 15 minutes, data were recorded.ResultsA significant decrease in BIS values (P < 0.01) associated with a position change from neutral position to beach chair position was evident.ConclusionsBIS values are significantly decreased in the beach chair position compared with the neutral position and might affect interpretation of the depth of anesthesia.

In recent years, the incorporation of NFC (Near Field Communication) technology into mobile devices has led to changes in payment system environments. Currently, the NFC mobile payment service is leading the mobile payment market. In particular, most electronic payment services, such as those used by Google Inc. and Apple Inc., are adopting payment methods based on NFC to replace credit cards. In addition, related groups from the Republic of Korea have enhanced safety in communication by using standard techniques for activating NFC services. However, various security threats are still present in electronic payment methods that use NFC. In this paper, we propose a mutual authentication scheme based on a lattice for conditional anonymity in NFC-PCM (Near Field Communication-Passive Communication Mode) payment services environments.

BackgroundA correct estimate of the tracheal tube insertion depth can prevent complications, including endobronchial intubation and vocal cord trauma. We evaluated a new topographical method for endotracheal tube positioning relative to the carina, using a well-known prior topographical method for comparison.MethodsOne hundred adult (male 50, female 50) patients were studied. The comparison topographic length (in cm) was measured by adding the distance between the right mouth corner and the right mandibular angle to the distance between the right mandibular angle and the center of the sternal manubrium. The new endotracheal tube insertion depth (in cm) was determined by adding the distance between the right mouth corner and the vocal cords, measured with the endotracheal tube itself, to the distance between the thyroid prominence and the manubriosternal joint, and then subtracting 4 cm. After intubation, the endotracheal tube was positioned properly at the right mouth corner and the endotracheal tube tip was evaluated using a fiberoptic bronchoscope at the carina.ResultsThe distances from the tip of the endotracheal tube to the carina were not significantly different between the methods in the same gender. However, our method allowed endotracheal tube tip placement between 3 cm and 5 cm, above the carina more frequently than the prior method in males.ConclusionsThe new topographical method can be used as a guide to positioning the endotracheal tubes.

BackgroundDuring neurosurgical procedures, patients are often exposed to hypoxic and ischemic brain damage. Cerebral ischemia leads to neuronal cell death and eventually causes neurological impairments. Remifentanil is a new ultra-short acting phenylpiperidine opioid analgesic. In this study, we evaluated remifentanil to determine if it exerts an anti-apoptotic effect in the hippocampal dentate gyrus following transient global ischemia in gerbils.MethodsStep-down avoidance task, terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assay, and immunohistochemical staining for caspase-3 were performed.ResultsThe numbers of TUNEL-positive cells and caspase-3-positive cells in the dentate gyrus were increased by ransient global ischemia. Latency in the step-down avoidance task was increased by transient global ischemia. Results revealed that apoptotic cell death in the dentate gyrus was increased significantly following transient global ischemia, resulting in memory impairment. However, treatment with remifentanil suppressed ischemia-induced apoptosis in the dentate gyrus, thereby alleviating the memory impairment that was induced by ischemic cerebral injury.ConclusionsThese results indicate that remifentanil may exert a neuroprotective effect on ischemic brain damage during surgery.

Introduction: A change from the supine to prone position causes hemodynamic alterations. We aimed to evaluate the effect of fluid preloading in the supine position, the subsequent hemodynamic changes in the prone position and postoperative outcomes. Patients and methods: This prospective, assessor-blind, randomized controlled trial was conducted between March and June 2023. Adults scheduled for elective orthopaedic lumbar surgery under general anaesthesia were enrolled. In total, 80 participants were randomly assigned to fluid maintenance (M) or loading (L) groups. Both groups were administered intravenous fluid at a rate of 2 ml/kg/h until surgical incision; Group L was loaded with an additional 5 ml/kg intravenous fluid for 10 min after anaesthesia induction. The primary outcome was incidence of hypotension before surgical incision. Secondary outcomes included differences in the mean blood pressure (mBP), heart rate, pleth variability index (PVi), stroke volume variation (SVV), pulse pressure variation (PPV), stroke volume index and cardiac index before surgical incision between the two groups. Additionally, postoperative complications until postoperative day 2 and postoperative hospital length of stay were investigated. Results: Hypotension was prevalent in Group M before surgical incision and could be predicted by a baseline PVi >16. The mBP was significantly higher in Group L immediately after fluid loading. The PVi, SVV and PPV were lower in Group L after fluid loading, with continued differences at 2-3 time points for SVV and PPV. Other outcomes did not differ between the two groups. Conclusion: Fluid loading after inducing general anaesthesia could reduce the occurrence of hypotension until surgical incision in patients scheduled for surgery in the prone position. Additionally, hypotension could be predicted in patients with a baseline PVi >16. Therefore, intravenous fluid loading is strongly recommended in patients with high baseline PVi to prevent hypotension after anaesthesia induction and in the prone position. Trial number: KCT0008294 (date of registration: 16 March 2023) [ABSTRACT FROM AUTHOR]