• Korean Journal of Radiology
• /
• v.23 no.10
• /
• pp.949-958
• /
• 2022

Objective: To investigate the feasibility of using a deep learning-based analysis of auscultation data to predict significant stenosis of arteriovenous fistulas (AVF) in patients undergoing hemodialysis requiring percutaneous transluminal angioplasty (PTA). Materials and Methods: Forty patients (24 male and 16 female; median age, 62.5 years) with dysfunctional native AVF were prospectively recruited. Digital sounds from the AVF shunt were recorded using a wireless electronic stethoscope before (pre-PTA) and after PTA (post-PTA), and the audio files were subsequently converted to mel spectrograms, which were used to construct various deep convolutional neural network (DCNN) models (DenseNet201, EfficientNetB5, and ResNet50). The performance of these models for diagnosing ≥ 50% AVF stenosis was assessed and compared. The ground truth for the presence of ≥ 50% AVF stenosis was obtained using digital subtraction angiography. Gradient-weighted class activation mapping (Grad-CAM) was used to produce visual explanations for DCNN model decisions. Results: Eighty audio files were obtained from the 40 recruited patients and pooled for the study. Mel spectrograms of "pre-PTA" shunt sounds showed patterns corresponding to abnormal high-pitched bruits with systolic accentuation observed in patients with stenotic AVF. The ResNet50 and EfficientNetB5 models yielded an area under the receiver operating characteristic curve of 0.99 and 0.98, respectively, at optimized epochs for predicting ≥ 50% AVF stenosis. However, Grad-CAM heatmaps revealed that only ResNet50 highlighted areas relevant to AVF stenosis in the mel spectrogram. Conclusion: Mel spectrogram-based DCNN models, particularly ResNet50, successfully predicted the presence of significant AVF stenosis requiring PTA in this feasibility study and may potentially be used in AVF surveillance.

• Information Systems Review
• /
• v.20 no.2
• /
• pp.139-161
• /
• 2018

As Korea became one of the oldest countries in the world, successful aging emerged as an important issue to individuals as well as to society. This study aims to determine not only the Korean seniors' happiness and unhappiness factors but also the means to enhance their happiness and deal with unhappiness. We collected news articles related to the happiness and unhappiness of seniors with nine keywords based on Alderfer's ERG Theory. We then applied a topic modeling technique, Latent Dirichlet Allocation, to examine the main issues underlying the seniors' happiness and unhappiness. According to the analysis, we investigated the conditions of happiness and unhappiness by inspecting the topics based on each keyword. We also conducted a detailed analysis based on the main factors from topic modeling. We proposed specific ways to increase and overcome the happiness and unhappiness of seniors, respectively, in terms of government, corporate, family, and other social welfare organizations. This study indicates the major factors that affect the happiness and unhappiness of seniors. Specific methods to boost happiness and relief unhappiness are suggested from the additional analysis.

• IMMUNE NETWORK
• /
• v.21 no.4
• /
• pp.25.1-25.16
• /
• 2021

Asthma is a heterogeneous disease whose development is shaped by a variety of environmental and genetic factors. While several recent studies suggest that microbial dysbiosis in the gut may promote asthma, little is known about the relationship between the recently discovered lung microbiome and asthma. Innate lymphoid cells (ILCs) have also been shown recently to participate in asthma. To investigate the relationship between the lung microbiome, ILCs, and asthma, we recruited 23 healthy controls (HC), 42 patients with non-severe asthma, and 32 patients with severe asthma. Flow cytometry analysis showed severe asthma associated with fewer natural cytotoxicity receptor (NCR)⁺ILC3s in the lung. Similar changes in other ILC subsets, macrophages, and monocytes were not observed. The asthma patients did not differ from the HC in terms of the alpha and beta-diversity of the lung and gut microbiomes. However, lung function correlated positively with both NCR⁺ILC3 frequencies and microbial diversity in the lung. Sputum NCR⁺ILC3 frequencies correlated positively with lung microbiome diversity in the HC, but this relationship was inversed in severe asthma. Together, these data suggest that airway NCR⁺ILC3s may contribute to a healthy commensal diversity and normal lung function.

• IMMUNE NETWORK
• /
• v.21 no.4
• /
• pp.29.1-29.9
• /
• 2021

There are limited data directly comparing humoral and T cell responses to the ChAdOx1 nCoV-19 and BNT162b2 vaccines. We compared Ab and T cell responses after first doses of ChAdOx1 nCoV-19 vs. BNT162b2 vaccines. We enrolled healthcare workers who received ChAdOx1 nCoV-19 or BNT162b2 vaccine in Seoul, Korea. Anti-severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) S1 protein-specific IgG Abs (S1-IgG), neutralizing Abs (NT Abs), and SARS-CoV-2-specific T cell response were evaluated before vaccination and at 1-wk intervals for 3 wks after vaccination. A total of 76 persons, comprising 40 injected with the ChAdOx1 vaccine and 36 injected with the BNT162b2 vaccine, participated in this study. At 3 wks after vaccination, the mean levels (±SD) of S1-IgG and NT Abs in the BNT162b2 participants were significantly higher than in the ChAdOx1 participants (S1-IgG, 14.03±7.20 vs. 6.28±8.87, p<0.0001; NT Ab, 183.1±155.6 vs. 116.6±116.2, p=0.035), respectively. However, the mean values of the T cell responses in the 2 groups were comparable after 2 wks. The humoral immune response after the 1st dose of BNT162b2 developed faster and was stronger than after the 1st dose of ChAdOx1. However, the T cell responses to BNT162b2 and ChAdOx1 were similar.

• IMMUNE NETWORK
• /
• v.22 no.2
• /
• pp.17.1-17.14
• /
• 2022

We aimed to investigate associations of dietary diversity (DD) with gut microbial diversity and the development of hen's egg allergy (HEA) in infants. We enrolled 68 infants in a high-risk group and 32 infants in a control group based on a family history of allergic diseases. All infants were followed from birth until 12 months of age. We collected infant feeding data, and DD was defined using 3 measures: the World Health Organization definition of minimum DD, food group diversity, and food allergen diversity. Gut microbiome profiles and expression of cytokines were evaluated by bacterial 16S rRNA sequencing and real-time reverse transcriptase-polymerase chain reaction. High DD scores at 3 and 4 months were associated with a lower risk of developing HEA in the high-risk group, but not in the control group. In the high-risk group, high DD scores at 3, 4, and 5 months of age were associated with an increase in Chao1 index at 6 months. We found that the gene expression of IL-4, IL-5, IL-6, and IL-8 were higher among infants who had lower DD scores compared to those who had higher DD scores in high-risk infants. Additionally, high-risk infants with a higher FAD score at 5 months of age showed a reduced gene expression of IL-13. Increasing DD within 6 months of life may increase gut microbial diversity, and thus reduce the development of HEA in infants with a family history of allergic diseases.

• IMMUNE NETWORK
• /
• v.22 no.3
• /
• pp.24.1-24.12
• /
• 2022

Coronavirus disease 2019 (COVID-19) vaccination in immunocompromised, especially transplant recipients, may induce a weaker immune response. But there are limited data on the immune response after COVID-19 vaccination in liver transplant (LT) recipients, especially on the comparison of Ab responses after different vaccine platforms between mRNA and adenoviral vector vaccines. Thus, we conducted a prospective study on LT recipients who received two doses of the ChAdOx1 nCoV-19 (ChAdOx1), mRNA-1273, or BNT162b2 vaccines compared with healthy healthcare workers (HCWs). SARS-CoV-2 S1-specific IgG Ab titers were measured using ELISA. Overall, 89 LT recipients (ChAdOx1, n=16 [18%]) or mRNA vaccines (mRNA-1273 vaccine, n=23 [26%]; BNT162b2 vaccine, n=50 [56%]) received 3 different vaccines. Of them, 16 (18%) had a positive Ab response after one dose of COVID-19 vaccine and 62 (73%) after 2 doses. However, the median Ab titer after two doses of mRNA vaccines was significantly higher (44.6 IU/ml) than after two doses of ChAdOx1 (19.2 IU/ml, p=0.04). The longer time interval from transplantation was significantly associated with high Ab titers after two doses of vaccine (p=0.003). However, mycophenolic acid use was not associated with Ab titers (p=0.53). In conclusion, about 3-quarters of LT recipients had a positive Ab response after 2 doses of vaccine, and the mRNA vaccines induced higher Ab responses than the ChAdOx1 vaccine.

• IMMUNE NETWORK
• /
• v.20 no.5
• /
• pp.43.1-43.11
• /
• 2020

Capicua (CIC) is a transcriptional repressor that regulates several developmental processes. CIC deficiency results in lymphoproliferative autoimmunity accompanied by expansion of CD44^hiCD62L^lo effector/memory and follicular Th cell populations. Deletion of Cic alleles in hematopoietic stem cells (Vav1-Cre-mediated knockout of Cic) causes more severe autoimmunity than that caused by the knockout of Cic in CD4⁺CD8⁺ double positive thymocytes (Cd4-Cre-mediated knockout of Cic). In this study, we compared splenic CD4⁺ T cell activation and proliferation between whole immune cell-specific Cic-null (Cic^f/f;Vav1-Cre) and T cell-specific Cic-null (Cic^f/f;Cd4-Cre) mice. Hyperactivation and hyperproliferation of CD4⁺ T cells were more apparent in Cic^f/f;Vav1-Cre mice than in Cic^f/f;Cd4-Cre mice. Cic^f/f;Vav1-Cre CD4⁺ T cells more rapidly proliferated and secreted larger amounts of IL-2 upon TCR stimulation than did Cic^f/f;Cd4-Cre CD4⁺ T cells, while the TCR stimulation-induced activation of the TCR signaling cascade and calcium flux were comparable between them. Mixed wild-type and Cic^f/f;Vav1-Cre bone marrow chimeras also exhibited more apparent hyperactivation and hyperproliferation of Cic-deficient CD4⁺ T cells than did mixed wild-type and Cic^f/f;Cd4-Cre bone marrow chimeras. Taken together, our data demonstrate that CIC deficiency at the beginning of T cell development endows peripheral CD4⁺ T cells with enhanced T cell activation and proliferative capability.

• IMMUNE NETWORK
• /
• v.20 no.5
• /
• pp.41.1-41.11
• /
• 2020

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV2) is a positive-sense single-stranded RNA (+ssRNA) that causes coronavirus disease 2019 (COVID-19). The viral genome encodes twelve genes for viral replication and infection. The third open reading frame is the spike (S) gene that encodes for the spike glycoprotein interacting with specific cell surface receptor - angiotensin converting enzyme 2 (ACE2) - on the host cell membrane. Most recent studies identified a single point mutation in S gene. A single point mutation in S gene leading to an amino acid substitution at codon 614 from an aspartic acid 614 into glycine (D614G) resulted in greater infectivity compared to the wild type SARS-CoV2. We were interested in investigating the mutation region of S gene of SARS-CoV2 from Korean COVID-19 patients. New mutation sites were found in the critical receptor binding domain (RBD) of S gene, which is adjacent to the aforementioned D614G mutation residue. This specific sequence data demonstrated the active progression of SARS-CoV2 by mutations in the RBD of S gene. The sequence information of new mutations is critical to the development of recombinant SARS-CoV2 spike antigens, which may be required to improve and advance the strategy against a wide range of possible SARS-CoV2 mutations.

• Informatization Policy
• /
• v.31 no.2
• /
• pp.3-38
• /
• 2024

Hallucination is a significant barrier to the utilization of large-scale language models or multimodal models. In this study, we collected 654 computer science papers with "hallucination" in the abstract from arXiv from December 2022 to January 2024 following the advent of Chat GPT and conducted frequency analysis, knowledge network analysis, and literature review to explore the latest trends in hallucination research. The results showed that research in the fields of "Computation and Language," "Artificial Intelligence," "Computer Vision and Pattern Recognition," and "Machine Learning" were active. We then analyzed the research trends in the four major fields by focusing on the main authors and dividing them into data, hallucination detection, and hallucination mitigation. The main research trends included hallucination mitigation through supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF), inference enhancement via "chain of thought" (CoT), and growing interest in hallucination mitigation within the domain of multimodal AI. This study provides insights into the latest developments in hallucination research through a technology-oriented literature review. This study is expected to help subsequent research in both engineering and humanities and social sciences fields by understanding the latest trends in hallucination research.

• Journal of Korea Water Resources Association
• /
• v.57 no.2
• /
• pp.73-85
• /
• 2024

Predicting water quality of rivers and reservoirs is necessary for the management of water resources. Artificial Neural Networks (ANNs) have been used in many studies to predict water quality with high accuracy. Previous studies have used Gradient Descent (GD)-based optimizers as an optimizer, an operator of ANN that searches parameters. However, GD-based optimizers have the disadvantages of the possibility of local optimal convergence and absence of a solution storage and comparison structure. This study developed improved optimizers to overcome the disadvantages of GD-based optimizers. Proposed optimizers are optimizers that combine adaptive moments (Adam) and Nesterov-accelerated adaptive moments (Nadam), which have low learning errors among GD-based optimizers, with Harmony Search (HS) or Novel Self-adaptive Harmony Search (NSHS). To evaluate the performance of Long Short-Term Memory (LSTM) using improved optimizers, the water quality data from the Dasan water quality monitoring station were used for training and prediction. Comparing the learning results, Mean Squared Error (MSE) of LSTM using Nadam combined with NSHS (NadamNSHS) was the lowest at 0.002921. In addition, the prediction rankings according to MSE and R² for the four water quality indices for each optimizer were compared. Comparing the average of ranking for each optimizer, it was confirmed that LSTM using NadamNSHS was the highest at 2.25.