A5064595835- Topic Modeling
- Machine Learning in Healthcare
- Neuroscience and Neuropharmacology Research
- Ion channel regulation and function
- Neurotransmitter Receptor Influence on Behavior
- Natural Language Processing Techniques
- Receptor Mechanisms and Signaling
- Artificial Intelligence in Healthcare
- Speech and dialogue systems
- Neuroscience and Neural Engineering
- Biomedical Text Mining and Ontologies
- Neural dynamics and brain function
- Evacuation and Crowd Dynamics
- Cellular transport and secretion
- Environmental Changes in China
- Time Series Analysis and Forecasting
- Flood Risk Assessment and Management
- Cell Adhesion Molecules Research
- Skin and Cellular Biology Research
- Disaster Management and Resilience
- Wheat and Barley Genetics and Pathology
- Diet and metabolism studies
- Crop Yield and Soil Fertility
- Cellular Mechanics and Interactions
- Sleep and Wakefulness Research
University of Chinese Academy of Sciences
2024
Institute of Software
2023-2024
Peking University
2017-2024
University of the Western Cape
2024
Institute of Genetics and Developmental Biology
2024
Chinese Academy of Sciences
2024
Chinese Institute for Brain Research
2017-2022
Center for Life Sciences
2017-2022
University of Michigan
2003
Thomas Jefferson University
1997
Junctional epidermolysis bullosa with congenital pyloric or duodenal atresia is a distinct variant within this group of autosomal recessive blistering skin diseases. In study we demonstrate, for the first time, homozygous mutation in α6 integrin gene (ITGA6) family three affected individuals. For purpose, determined genomic organization ITGA6, and placed on chromosome 2q by high resolution radiation hybrid mapping. Heteroduplex analysis PCR products containing individual exons followed...
Deep learning techniques are capable of capturing complex input-output relationships, and have been widely applied to the diagnosis prediction task based on web-based patient electronic health records (EHR) data. To improve interpretability pure data-driven deep with only a limited amount labeled data, pervasive trend is assist model training knowledge priors from online medical graphs. However, they marginally investigated label imbalance task-irrelevant noise in external graph. The...
The norepinephrine neurons in locus coeruleus (LC-NE neurons) are essential for sleep arousal, pain sensation, and cocaine addiction. According to previous studies, increases NE overflow (the profile of extracellular level response stimulation) by blocking the reuptake. is determined release via exocytosis reuptake through transporter (NET). However, whether directly affects vesicular has not been tested. By recording quantal from LC-NE neurons, we report that frequency regulation NET...
Due to the insufficiency of electronic health records (EHR) data utilized in practical diagnosis prediction scenarios, most works are devoted learning powerful patient representations either from structured EHR (e.g., temporal medical events, lab test results, etc.) or unstructured clinical notes, etc.). However, synthesizing rich information both them still needs be explored. Firstly, heterogeneous semantic biases across heavily hinder synthesis representation spaces, which is critical for...
We explore how the rise of Large Language Models (LLMs) significantly impacts task performance in field Natural Processing. focus on two strategies, Retrieval-Augmented Generation (RAG) and Fine-Tuning (FT), propose Hypothesis Knowledge Graph Enhanced (HyKGE) framework, leveraging a knowledge graph to enhance medical LLMs. By integrating LLMs graphs, HyKGE demonstrates superior addressing accuracy interpretability challenges, presenting potential applications domain. Our evaluations using...
Graph Neural Networks (GNNs) have shown considerable effectiveness in a variety of graph learning tasks, particularly those based on the message-passing approach recent years. However, their performance is often constrained by limited receptive field, challenge that becomes more acute presence sparse graphs. In light power series, which possesses infinite expansion capabilities, we propose novel Power Filter Network (GPFN) enhances node classification employing series filter to augment...
With the widespread adoption of electronic health records (EHR) data, deep learning techniques have been broadly utilized for various prediction tasks. Nevertheless, labeled data scarcity issue restricts power these models. To enhance generalization capability models when faced with such situations, a common trend is to train generative adversarial networks (GANs) or diffusion augmentation. However, due limitations in sample size and potential label imbalance issues, methods are prone mode...
By integrating external knowledge, Retrieval-Augmented Generation (RAG) has become an effective strategy for mitigating the hallucination problems that large language models (LLMs) encounter when dealing with knowledge-intensive tasks. However, in process of non-parametric supporting evidence internal parametric inevitable knowledge conflicts may arise, leading to confusion model's responses. To enhance selection LLMs various contexts, some research focused on refining their behavior...
While pioneering deep learning methods have made great strides in analyzing electronic health record (EHR) data, they often struggle to fully capture the semantics of diverse medical codes from limited data. The integration external knowledge Large Language Models (LLMs) presents a promising avenue for improving healthcare predictions. However, LLM analyses may exhibit significant variance due ambiguity problems and inconsistency issues, hindering their effective utilization. To address...
Retrieval-Augmented Generation (RAG) offers an effective solution to the issues faced by Large Language Models (LLMs) in hallucination generation and knowledge obsolescence incorporating externally retrieved knowledge. However, due potential conflicts between internal external knowledge, as well retrieval noise, LLMs often struggle effectively integrate evidence, leading a decline performance. Although existing methods attempt tackle these challenges, they strike balance model adherence...
Large Language Models(LLMs) excel in general tasks but struggle specialized domains like healthcare due to limited domain-specific knowledge.Supervised Fine-Tuning(SFT) data construction for domain adaptation often relies on heuristic methods, such as GPT-4 annotation or manual selection, with a data-centric focus presumed diverse, high-quality datasets. However, these methods overlook the model's inherent knowledge distribution, introducing noise, redundancy, and irrelevant data, leading...
The catechol-O-methyl-transferase (COMT) gene regulates the metabolic processes of neurotransmitter dopamine and, through it, influences endorphins, which play an important role in process pain perception. It was found that COMT with amino acid valine (val158) is more active than variant containing methionine (met158).
As the population of women entering menopause grows, so does potential for chronic conditions like os-teoporosis, cardiovascular disease, and metabolic syndrome, which underscores critical need early intervention treatment during menopause. Meanwhile Electronic Healthcare Records (EHR) is becoming a valuable resource health event predictions, including diagnosis, mortality, length-of-stay, readmission. However, effectively utilizing diagnosis features co-occurrence relationships among...
Abstract A central principle of synaptic transmission is that action potential-induced presynaptic neurotransmitter release occurs exclusively via Ca 2+ -dependent secretion (CDS). The discovery and mechanistic investigations -independent but voltage-dependent (CiVDS) have demonstrated the potential per se sufficient to trigger neurotransmission in somata primary sensory sympathetic neurons mammals. One key question remains, however, whether CiVDS contributes transmission. Here we report,...
SummaryThe norepinephrine neurons in locus coeruleus (LC-NE neurons) play key roles sleep-arousal, pain sensation, and cocaine addiction. According to the 50-year-old dogma, increases NE overflow by blocking reuptake. is essentially determined release reuptake through transporter. However, whether or not directly affects vesicular has never been tested due lack of a direct assay. By recording quantal from LC-NE neurons, we report that frequency transporter downstream PKC signaling;...