A5051404242- Cell Image Analysis Techniques
- Digital Holography and Microscopy
- Image Processing Techniques and Applications
- Monoclonal and Polyclonal Antibodies Research
- Advanced Biosensing Techniques and Applications
- Health, Environment, Cognitive Aging
- Homelessness and Social Issues
- Prenatal Substance Exposure Effects
- Gestational Diabetes Research and Management
- CAR-T cell therapy research
- Dementia and Cognitive Impairment Research
- Neural Networks and Applications
- Fire Detection and Safety Systems
- Nanofabrication and Lithography Techniques
- Bacterial Identification and Susceptibility Testing
- Advanced Statistical Methods and Models
- Immunotherapy and Immune Responses
- Liver Disease Diagnosis and Treatment
- Engineering Applied Research
- Photoacoustic and Ultrasonic Imaging
- Liver physiology and pathology
- Image and Signal Denoising Methods
- Ubiquitin and proteasome pathways
- Epigenetics and DNA Methylation
- Genetic Syndromes and Imprinting
Sungkyunkwan University
2025
Massachusetts Institute of Technology
2019-2024
Moscow Institute of Thermal Technology
2024
Korea Advanced Institute of Science and Technology
2003-2023
Kootenay Association for Science & Technology
2019
Kyungsung University
2014
The healthcare industry is in dire need of rapid microbial identification techniques for treating infections. Microbial infections are a major issue worldwide, as these widespread diseases often develop into deadly symptoms. While studies have shown that an early appropriate antibiotic treatment significantly reduces the mortality infection, this effective difficult to practice. main obstacle treatments long turnaround time routine identification, which includes time-consuming sample growth....
The immunological synapse (IS) is a cell-cell junction between T cell and professional antigen-presenting cell. Since the IS formation critical step for initiation of an antigen-specific immune response, various live-cell imaging techniques, most which rely on fluorescence microscopy, have been used to study dynamics IS. However, inherent limitations associated with fluorescence-based imaging, such as photo-bleaching photo-toxicity, prevent long-term assessment dynamic changes high...
Given the global increase in obesity, metabolic dysfunction-associated steatotic liver disease (MASLD) is a major health concern. Because primary organ for xenobiotic metabolism, impact of environmental stressors on homeostasis and MASLD has garnered significant interest over past few decades. The concept metabolism-disrupting chemicals (MDCs) been introduced to underscore importance factors homeostasis. Recent epidemiological biological studies suggest causal link between exposure MDCs...
Abstract DNA methylation serves as a powerful biomarker for disease diagnosis and biological age assessment. However, current analytical approaches often rely on linear models that cannot capture the complex, context-dependent nature of regulation. Here we present MethylGPT, transformer-based foundation model trained 226,555 (154,063 after QC deduplication) human profiles spanning diverse tissue types from 5,281 datasets, curated 49,156 CpG sites, 7.6 billion training tokens. MethylGPT...
Background Fetal alcohol syndrome (FAS) is a lifelong developmental disability that occurs among individuals with prenatal exposure (PAE). With improved prediction models, FAS can be diagnosed or treated early, if not completely prevented. Objective In this study, we sought to compare different machine learning algorithms and their predictive performance women who consumed during pregnancy. We also aimed identify which variables (eg, timing of pregnancy type consumed) were most influential...
Abstract The healthcare industry is in dire need for rapid microbial identification techniques. Microbial infection a major issue with significant prevalence and mortality, which can be treated effectively during the early stages using appropriate antibiotics. However, determining antibiotics treatment of remains challenge, mainly due to lack Conventional culture-based matrix-assisted laser desorption/ionization time-of-flight mass spectroscopy are gold standard methods, but sample...
We propose and experimentally validate a label-free, volumetric, automated assessment method of immunological synapse dynamics using combinational approach optical diffraction tomography deep learning-based segmentation. The proposed enables automatic quantitative spatiotemporal analyses kinetics regarding morphological biochemical parameters related to the total protein densities immune cells, thus providing new perspective for studies in immunology.
Neural scaling laws characterize how model performance improves as the size scales up. Inspired by empirical observations, we introduce a resource of neural scaling. A task is usually composite hence can be decomposed into many subtasks, which compete for resources (measured number neurons allocated to subtasks). On toy problems, empirically find that: (1) The loss subtask inversely proportional its neurons. (2) When multiple subtasks are present in task, acquired each uniformly grow models...
Background: Post-traumatic stress symptoms (PTSS) can emerge either immediately in the aftermath of disaster or with delayed onset, posing challenges due to their unpredictable nature and potential remain untreated for extended periods. Machine learning modeling may offer new insights predicting delayed-onset PTSS, thus filling a crucial gap our understanding post-disaster mental health trajectories.Methods: This study utilized prospective cohort Japanese older adults aged ≥65 years who were...
Kaplan et al. [2020] (`Kaplan') and Hoffmann [2022] (`Chinchilla') studied the scaling behavior of transformers trained on next-token language prediction. These studies produced different estimates for how number parameters ($N$) training tokens ($D$) should be set to achieve lowest possible loss a given compute budget ($C$). Kaplan: $N_\text{optimal} \propto C^{0.73}$, Chinchilla: C^{0.50}$. This note finds that much this discrepancy can attributed counting non-embedding rather than total...
<sec> <title>BACKGROUND</title> Mild cognitive impairment (MCI) poses significant challenges in early diagnosis and timely intervention. Underdiagnosis, coupled with the economic social burden of dementia, necessitates more precise detection methods. Machine learning (ML) algorithms show promise managing complex data for MCI dementia prediction. </sec> <title>OBJECTIVE</title> This study assessed predictive accuracy ML models identifying onset using Korean Longitudinal Study Aging (KLoSA)...
Background Mild cognitive impairment (MCI) poses significant challenges in early diagnosis and timely intervention. Underdiagnosis, coupled with the economic social burden of dementia, necessitates more precise detection methods. Machine learning (ML) algorithms show promise managing complex data for MCI dementia prediction. Objective This study assessed predictive accuracy ML models identifying onset using Korean Longitudinal Study Aging (KLoSA) dataset. Methods used from KLoSA, a...
Humans distill complex experiences into fundamental abstractions that enable rapid learning and adaptation. Similarly, autoregressive transformers exhibit adaptive through in-context (ICL), which begs the question of how. In this paper, we propose \textbf{concept encoding-decoding mechanism} to explain ICL by studying how form use internal in their representations. On synthetic tasks, analyze training dynamics a small transformer report coupled emergence concept encoding decoding. As model...
Immunoglobulin G (IgG) antibodies are widely used for diagnosis and therapy. Given the unique dimeric structure of IgG, we hypothesized that, by genetically fusing a homodimeric protein (catenator) to C-terminus reversible catenation antibody molecules could be induced on surface where target antigen abundant, that it an effective way greatly enhance antigen-binding avidity. A thermodynamic simulation showed quite low homodimerization affinity catenator, e.g . dissociation constant 100 μM,...
A linear combination of weighted order statistic (LWOS) filters, which can be thought as an extension stack represent any Boolean function (BF) or its extension, is called the extended BF (EBF). The authors present a procedure for finding LWOS filter simplest type from filters are equivalent to given EBF. In addition, property that useful implementing derived and algorithm filtering presented.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
Rapid identification of infectious pathogens can save lives and mitigate healthcare expenses. Yet the current turnaround time for microbial typically exceeds 24 hours, as common methods require cultivation millions or more bacteria to detect collective signal. In this study, we propose a hybrid framework quantitative phase imaging artificial neural network facilitate rapid at an individual-cell level. Specifically, three-dimensional images refractive index were acquired individual bacteria,...
Rapid, label-free, volumetric, and automated assessment in microscopy is necessary to assess the dynamic interactions between lymphocytes their targets through immunological synapse (IS) relevant functions. However, attempts realize automatic tracking of IS dynamics have been stymied by limitations imaging techniques computational analysis methods. Here, we demonstrate three-dimensional combining optical diffraction tomography deep-learning-based segmentation. The proposed approach enables...
ABSTRACT Immunoglobulin G (IgG) antibodies are widely used for diagnosis and therapy. Given the unique dimeric structure of IgG, we hypothesized that, by genetically fusing a homodimeric protein (catenator) to C-terminus reversible catenation antibody molecules could be induced on surface where target antigen abundant, that it an effective way greatly enhance antigen-binding avidity. A thermodynamic simulation shows quite low homodimerization affinity catenator, e.g. dissociation constant...