A5001814373- Multimodal Machine Learning Applications
- Radiomics and Machine Learning in Medical Imaging
- Electrodeposition and Electroless Coatings
- COVID-19 diagnosis using AI
- Domain Adaptation and Few-Shot Learning
- Lung Cancer Diagnosis and Treatment
- Biomedical Text Mining and Ontologies
- Electrochemical Analysis and Applications
- Topic Modeling
- Neutrophil, Myeloperoxidase and Oxidative Mechanisms
- Advanced Image and Video Retrieval Techniques
- Human Pose and Action Recognition
- Bioinformatics and Genomic Networks
- Conferences and Exhibitions Management
- Immune cells in cancer
- Machine Learning in Bioinformatics
- Computational Drug Discovery Methods
- Cancer-related molecular mechanisms research
- Extraction and Separation Processes
- Data Quality and Management
- scientometrics and bibliometrics research
- Electronic Packaging and Soldering Technologies
- Semantic Web and Ontologies
- Misinformation and Its Impacts
- Cell Adhesion Molecules Research
Taiyuan University of Technology
2024
The University of Texas at Austin
2019-2023
Second Affiliated Hospital of Harbin Medical University
2021-2022
Harbin Medical University
2012-2022
North University of China
2019-2021
PubMed
Visual question answering is the task of questions about images. We introduce VizWiz-VQA-Grounding dataset, first dataset that visually grounds answers to visual asked by people with impairments. analyze our and compare it five VQA-Grounding datasets demonstrate what makes similar different. then evaluate SOTA VQA models current algorithms often fail identify correct evidence where answer located. These regularly struggle when occupies a small fraction image, for images are higher quality,...
Chest X-ray becomes one of the most common medical diagnoses due to its noninvasiveness. The number chest images has skyrocketed, but reading X-rays still been manually performed by radiologists, which creates huge burnouts and delays. Traditionally, radiomics, as a subfield radiology that can extract large quantitative features from images, demonstrates potential facilitate imaging diagnosis before deep learning era. With rise learning, explainability neural networks on remains opaque. In...
The development of venous thromboembolism (VTE) is associated with high mortality among gastric cancer (GC) patients. Neutrophil extracellular traps (NETs) have been reported to correlate the prothrombotic state in some diseases, but are rarely GC patients.To investigate effect NETs on cancer-associated thrombosis patients.The levels blood and tissue samples patients were analyzed by ELISA, flow cytometry, immunofluorescence staining. NET generation hypercoagulation platelets endothelial...
No existing work on visual question answering explicitly accounts for ambiguity regarding where the content described in is located image. To fill this gap, we introduce VQ-FocusAmbiguity, first VQA dataset that visually grounds each region necessary to arrive at answer. We then provide an analysis showing how our grounding `questions' distinct from `answers', and characterize properties of questions segmentations provided dataset. Finally, benchmark modern models two novel tasks:...
Scientific novelty drives the efforts to invent new vaccines and solutions during pandemic. First-time collaboration international are two pivotal channels expand teams' search activities for a broader scope of resources required address global challenge, which might facilitate generation novel ideas. Our analysis 98,981 coronavirus papers suggests that scientific measured by BioBERT model is pretrained on 29 million PubMed articles, first-time increased after outbreak COVID-19, witnessed...
Accurate classification and localization of abnormalities in chest X-rays play an important role clinical diagnosis treatment planning. Building a highly accurate predictive model for these tasks usually requires large number manually annotated labels pixel regions (bounding boxes) abnormalities. However, it is expensive to acquire such annotations, especially the bounding boxes. Recently, contrastive learning has shown strong promise leveraging unlabeled natural images produce generalizable...
Heart Rate Variability (HRV) is a significant indicator of physical fitness and mental stresses. Electrocardiography (ECG) the most common means to detect HRV. However, nature ECG detection limits its convenience usage time. While wrist-based Photoplethysmography (PPG) provides possibility convenient 24/7 HRV monitoring, accuracy has not been systematically analyzed. We experimented with six healthy subjects measured PPG from their wrist chest. The result shows that ten parameters have...
Background Gastric cancer (GC) is the fifth most common malignant tumor and third leading cause of cancer-related deaths worldwide. Neutrophil extracellular traps (NETs) can enhance invasion GC cells are associated with poor prognosis in patients. However, its mechanism action not completely understood. Methods The content NETs peripheral blood patients was detected by enzyme-linked immunosorbent assay. AGS were treated or without for 24 h. High-throughput RNA sequencing performed to screen...
The emergence of the novel COVID-19 pandemic has had a significant impact on global healthcare and economy over past few months. virus's rapid widespread led to proliferation in biomedical research addressing its related topics. One essential Knowledge Discovery tools that could help community understand eventually find cure for are Graphs. CORD-19 dataset is collection publicly available full-text articles have been recently published coronavirus Here, we use several Machine Learning, Deep...
Sn–SnS x is used as a bifunctional electrode in CO 2 RR//GOR coupling system, producing formate simultaneously at both the cathode and anode. This system combines two pathways “CO –HCOO − ” –C 6 H 12 O ”.
This study builds a coronavirus knowledge graph (KG) by merging two information sources. The first source is Analytical Graph (AG), which integrates more than 20 different public datasets related to drug discovery. second CORD-19, collection of published scientific articles COVID-19. We combined both chemo genomic entities in AG with extracted from CORD-19 expand the COVID-19 domain. Before populating KG those entities, we perform entity disambiguation on collections using Wikidata. Our...
Chest X-ray becomes one of the most common medical diagnoses due to its noninvasiveness. The number chest images has skyrocketed, but reading X-rays still have been manually performed by radiologists, which creates huge burnouts and delays. Traditionally, radiomics, as a subfield radiology that can extract large quantitative features from images, demonstrates potential facilitate imaging diagnosis before deep learning era. In this paper, we develop an end-to-end framework, ChexRadiNet,...
PubMed is an essential resource for the medical domain, but useful concepts are either difficult to extract or ambiguated, which has significantly hindered knowledge discovery. To address this issue, we constructed a graph (PKG) by extracting bio-entities from 29 million abstracts, disambiguating author names, integrating funding data through National Institutes of Health (NIH) ExPORTER, collecting affiliation history and educational background authors ORCID, identifying fine-grained...
A comprehensive evaluation is critical to assess the capabilities of large multimodal models (LMM). In this study, we evaluate state-of-the-art LMMs, namely GPT-4V and Gemini, utilizing VQAonline dataset. an end-to-end authentic VQA dataset sourced from a diverse range everyday users. Compared previous benchmarks, well aligns with real-world tasks. It enables us effectively generality LMM, facilitates direct comparison human performance. To comprehensively generate seven types metadata for...
Visual question answering is a task of predicting the answer to about an image. Given that different people can provide answers visual question, we aim better understand why with groundings. We introduce first dataset visually grounds each unique which call VQA-AnswerTherapy. then propose two novel problems whether has single grounding and localizing all benchmark modern algorithms for these show where they succeed struggle. The evaluation server be found publicly at...