A5060805888- Diabetes Management and Research
- Dysphagia Assessment and Management
- Tracheal and airway disorders
- ECG Monitoring and Analysis
- Space Exploration and Technology
- AI in cancer detection
- COVID-19 diagnosis using AI
- Diabetes and associated disorders
- Diabetes Treatment and Management
- Spaceflight effects on biology
- Cardiac Arrest and Resuscitation
- Animal testing and alternatives
- Aerodynamics and Acoustics in Jet Flows
- Sepsis Diagnosis and Treatment
- Healthcare Technology and Patient Monitoring
- Data Stream Mining Techniques
- Contact Dermatitis and Allergies
- Diabetes, Cardiovascular Risks, and Lipoproteins
- Heart Rate Variability and Autonomic Control
- Infection Control and Ventilation
- Blood Pressure and Hypertension Studies
- Hyperglycemia and glycemic control in critically ill and hospitalized patients
- Metabolomics and Mass Spectrometry Studies
- Radiomics and Machine Learning in Medical Imaging
- Machine Learning in Healthcare
Ulsan National Institute of Science and Technology
2024-2025
Seoul National University Hospital
2022-2024
Seoul National University
2015-2023
Background and Purpose— The aim of this study was to explore clinical radiological prognostic factors for long-term swallowing recovery in patients with poststroke dysphagia develop validate a model using machine learning algorithm. Methods— Consecutive (N=137) acute ischemic stroke referred examinations were retrospectively reviewed. Dysphagia monitored the 6 months period then analyzed Kaplan-Meier method Cox regression factors. Bayesian network models developed potential classify into...
Abstract Computer-aided detection (CADe) systems have been actively researched for polyp in colonoscopy. To be an effective system, it is important to detect additional polyps that may easily missed by endoscopists. Sessile serrated lesions (SSLs) are a precursor colorectal cancer with relatively higher miss rate, owing their flat and subtle morphology. Colonoscopy CADe could help endoscopists; however, the current exhibit very low performance detecting SSLs. We propose system reflects...
Abstract Gravitational forces can impose physical stresses on the human body as it functions to maintain homeostasis. It has been reported that astronauts exposed microgravity experience altered biological and many subsequent studies effects of have therefore conducted. However, anticancer mechanisms simulated remain unclear. We previously showed proliferation Hodgkin’s lymphoma (HL) cells was inhibited when these were cultured in time-averaged (taSMG). In present study, we investigated...
While continuous glucose monitoring (CGM) has revolutionized metabolic health management, widespread adoption remains limited by cost constraints and usage burden, often resulting in interrupted periods. We propose a deep learning framework for level inference that operates independently of prior measurements, utilizing comprehensive life-log data. The model employs bidirectional Long Short-Term Memory (LSTM) network with an encoder-decoder architecture, incorporating dual attention...
Abstract Exposure to microgravity affects human physiology in various ways, and astronauts frequently report skin-related problems. Skin rash irritation are frequent complaints during space missions, skin thinning has also been reported after returning Earth. However, spaceflight missions for studying the physiological changes impractical. Thus, we used a previously developed 3D clinostat simulate environment investigate whether of can be reproduced vitro setting. Our results showed that...
Controlling blood glucose levels in diabetic patients is important for managing their health and quality of life. Several algorithms based on model predictive control reinforcement learning (RL) have been proposed so far, most which use prior knowledge physiological systems, the mathematical structure dynamics, many episodes including failures training policy network RL. To be smoothly adopted clinical settings, we propose a fast online method underlining safety interpretability. A random...
Abstract Prediction of bacteremia is a clinically important but challenging task. An artificial intelligence (AI) model has the potential to facilitate early prediction, aiding emergency department (ED) physicians in making timely decisions and reducing unnecessary medical costs. In this study, we developed externally validated Bayesian neural network-based AI prediction (AI-BPM). We also evaluated its impact on physician predictive performance considering both uncertainties using historical...
Abstract Identification of prognostic factors for swallowing recovery in patients with post-stroke dysphagia is crucial determining therapeutic strategies. We aimed at exploring hyoid kinematic features poor prognosis dysphagia. Of 122 who experienced following ischemic stroke, 18 prognosis, and age- sex-matched good were selected retrospectively reviewed. Positional data the bone during obtained from initial videofluoroscopic study after stroke onset. Normalized profiles...
Deep learning has been increasingly utilized in the medical field and achieved many goals. Since size of data dominates performance deep learning, several institutions are conducting joint research to obtain as much possible. However, sharing is usually prohibited owing risk privacy invasion. Federated a reasonable idea train distributed multicenter without direct access; however, central server merge distribute models needed, which expensive hardly approved due various legal regulations....
Breast cancer is the second leading for Korean women and its incidence rate has been increasing annually. If early diagnosis were implemented with epidemiologic data, could easily assess breast risk using internet. National Cancer Institute in United States released a Web-based Risk Assessment Tool based on Gail model. However, it inapplicable directly to since dependent race. Also, shows low accuracy (58%-59%). In this study, discrimination models are developed only epidemiological...
Abstract Many defined approaches (DAs) for skin sensitization assessment based on the adverse outcome pathway (AOP) have been developed to replace animal testing because European Union has banned cosmetic ingredients. Several DAs demonstrated that machine learning models are beneficial. In this study, we an ensemble prediction model utilizing graph convolutional network (GCN) and approach assess sensitization. The integrates in silico parameters data from alternatives of well‐defined AOP...
On-scene resuscitation time is associated with out-of-hospital cardiac arrest (OHCA) outcomes. We developed and validated reinforcement learning models for individualized on-scene times, leveraging nationwide Korean data. Adult OHCA patients a medical cause of were included (N = 73,905). The optimal policy was derived from conservative Q-learning to maximize survival. return spontaneous circulation hazard rates estimated the Random Survival Forest used as intermediate rewards handle sparse...
The objective of this study is to propose MD-VAE: a multi-task disentangled variational autoencoders (VAE) for exploring characteristics latent representations (LR) and exploiting LR diverse tasks including glucose forecasting, event detection, temporal clustering. We applied MD-VAE one virtual continuous monitoring (CGM) data from an FDA-approved Type 1 Diabetes Mellitus simulator (T1DMS) publicly available CGM real patients dynamics Mellitus. captured meaningful information be exploited...
Aim To study the effects of angiotensin receptor blockers (ARBs) on insulin secretion in hypertensive patients with type 2 diabetes. Materials and methods A total 41 were enrolled this open‐label, active comparator‐controlled, crossover study. After a 2‐week run‐in period amlodipine, participants assigned to receive either fimasartan (60–120 mg daily) or amlodipine (5–10 for 16 weeks. Thereafter, they treated other drug another Physical examinations laboratory tests performed before after...
Background The oral minimal model is a simple, useful tool for the assessment of β-cell function and insulin sensitivity across spectrum glucose tolerance, including normal tolerance (NGT), prediabetes, type 2 diabetes mellitus (T2DM) in humans. Methods Plasma glucose, insulin, C-peptide levels were measured during 180-minute, 75-g test 24 Korean subjects with NGT (n=10) T2DM (n=14). parameters computational estimated, indexes compared between groups. Results index was lower group than...