检索结果-内蒙古大学图书馆

STASCAN deciphers fine-resolution cell distribution maps in spatial transcriptomics by deep learning

GENOME BIOLOGY 2024年第1期25卷 1-28页

作者： Wu, Ying Zhou, Jia-Yi Yao, Bofei Cui, Guanshen Zhao, Yong-Liang Gao, Chun-Chun Yang, Ying Zhang, Shihua Yang, Yun-Gui China Natl Ctr Bioinformat Beijing 100101 Peoples R China Chinese Acad Sci Beijing Inst Genom Beijing 100101 Peoples R China Univ Chinese Acad Sci Beijing 100049 Peoples R China Univ Chinese Acad Sci Coll Future Technol Beijing 100049 Peoples R China Chinese Acad Sci Acad Math & Syst Sci CEMS RCSDSNCMIS Beijing 100190 Peoples R China Univ Chinese Acad Sci Sch Math Sci Beijing 100049 Peoples R China Univ Chinese Acad Sci Hangzhou Inst Adv Study Sch Life Sci Key Lab Syst Hlth Sci Zhejiang Prov Hangzhou 310024 Peoples R China Univ Chinese Acad Sci Sino Danish Coll Beijing 101408 Peoples R China Chinese Acad Sci Inst Stem Cell & Regenerat Beijing 100101 Peoples R China

Spatial transcriptomics technologies have been widely applied to decode cellular distribution by resolving gene expression profiles in tissue. However, sequencing techniques still limit the ability to create a fine-resolved spatial cell-type map. To this end, we develop a novel deep-learning-based approach, STASCAN, to predict the spatial cellular distribution of captured or uncharted areas where only histology images are available by cell feature learning integrating gene expression profiles and histology images. STASCAN is successfully applied across diverse datasets from different spatial transcriptomics technologies and displays significant advantages in deciphering higher-resolution cellular distribution and resolving enhanced organizational structures.

关键词： Spatial transcriptomics Cell annotation Deep learning Imputation multimodal data integration

来源：评论

学校读者我要写书评

暂无评论

A novel deep learning algorithm for real-time prediction of clinical deterioration in the emergency department for a multimodal clinical decision support system

引用

SCIENTIFIC REPORTS 2024年第1期14卷 1-15页

作者： Choi, Arom Lee, Kwanhyung Hyun, Heejung Kim, Kwang Joon Ahn, Byungeun Lee, Kyung Hyun Hahn, Sangchul Choi, So Yeon Kim, Ji Hoon Yonsei Univ Dept Emergency Med Coll Med 50-1 Yonsei Ro Seoul South Korea Yonsei Univ Inst Innovat Digital Healthcare 50 Yonsei Ro Seoul South Korea AITRICS Corp 218 Teheran Ro Seoul South Korea Yonsei Univ Dept Internal Med Div Geriatr Coll Med Seoul South Korea

The array of complex and evolving patient data has limited clinical decision making in the emergency department (ED). This study introduces an advanced deep learning algorithm designed to enhance real-time prediction accuracy for integration into a novel Clinical Decision Support System (CDSS). A retrospective study was conducted using data from a level 1 tertiary hospital. The algorithm's predictive performance was evaluated based on in-hospital cardiac arrest, inotropic circulatory support, advanced airway, and intensive care unit admission. We developed an artificial intelligence (AI) algorithm for CDSS that integrates multiple data modalities, including vitals, laboratory, and imaging results from electronic health records. The AI model was trained and tested on a dataset of 237,059 ED visits. The algorithm's predictions, based solely on triage information, significantly outperformed traditional logistic regression models, with notable improvements in the area under the precision-recall curve (AUPRC). Additionally, predictive accuracy improved with the inclusion of continuous data input at shorter intervals. This study suggests the feasibility of using AI algorithms in diverse clinical scenarios, particularly for earlier detection of clinical deterioration. Future work should focus on expanding the dataset and enhancing real-time data integration across multiple centers to further optimize its application within the novel CDSS.

关键词： Clinical decision support system Emergency department Deep learning multimodal data integration Real-time prediction Patient deterioration

来源：评论

学校读者我要写书评

暂无评论

Classification of lumbar spine disorders using large language models and MRI segmentation

引用

BMC MEDICAL INFORMATICS AND DECISION MAKING 2024年第1期24卷 1-18页

作者： Dong, Rongpeng Cheng, Xueliang Kang, Mingyang Qu, Yang Second Hosp Jilin Univ Dept Spinal Surg 218 Ziqiang St Chuangchun 130041 Peoples R China

BackgroundMRI is critical for diagnosing lumbar spine disorders but its complexity challenges diagnostic accuracy. This study proposes a BERT-based large language model (LLM) to enhance precision in classifying lumbar spine disorders through the integration of MRI data, textual reports, and numerical *** segmentation quality of MRI data is evaluated using dice coefficients (cut-off: 0.92) and intersection over union (IoU) metrics (cut-off: 0.88) to ensure precise anatomical feature extraction. The CNN extracts key lumbar spine features, such as lumbar lordotic angle (LLA) and disc heights, which are tokenized as direct scalar values representing positional relationships. A data source of 28,065 patients with various disorders, including degenerative disc disease, spinal stenosis, and spondylolisthesis, is used to establish diagnostic standards. These standards are refined through post-CNN processing of MRI texture features. The BERT-based spinal LLM model integrates these CNN-extracted MRI features and numerical values through early fusion *** analysis illustrate various lumbar spine disorders and their anatomical changes. The model achieved high performance, with all key metrics nearing 0.9, demonstrating its effectiveness in classifying conditions like spondylolisthesis, herniated disc, and spinal stenosis. External validation further confirmed the model's generalizability across different populations. External validation on 514 expert-validated MRI cases further confirms the model's clinical relevance and generalizability. The BERT-based model classifies 61 combinations of lumbar spine *** BERT-based spinal LLM significantly improves the precision of lumbar spine disorder classification, supporting accurate diagnosis and treatment planning.

关键词： Lumbar spine disorders MRI data BERT-based large language model multimodal data integration Precision classification

来源：评论

学校读者我要写书评

暂无评论

Artificial intelligence in COVID-19 research: A comprehensive survey of innovations, challenges, and future directions

引用

COMPUTER SCIENCE REVIEW 2025年 57卷

作者： Annan, Richard Qingge, Letu North Carolina A&T State Univ Dept Comp Sci Greensboro NC 27411 USA

The COVID-19 pandemic has accelerated the use of AI and ML in healthcare, improving diagnosis, treatment, and resource allocation. This survey examines the AI applications in disease detection, differential diagnosis, and post-COVID complication analysis. Our findings show that 53% of the reviewed studies focus on COVID19 detection, while only 14% address post-COVID complications. This reveals a gap in long-term patient monitoring. Convolutional Neural Networks (CNNs) are the most frequently used models, appearing in 23% of reviewed studies as standalone architectures and even more often in hybrid models. Meanwhile, transformers and multimodal models remain underutilized. Each appears in only 4% of the studies, limiting the integration of diverse data sources, such as imaging, audio, and lab results. Federated learning, a privacy-preserving AI approach, appears in 9% of studies. However, it is still less common than centralized models. This restricts secure and collaborative AI development. Despite these progress, challenges such as data bias, limited model generalization, and ethical concerns persist. Advanced methods including transformer models and knowledge distillation offer potential solutions for improving computational efficiency. To strengthen AI-driven healthcare, this survey highlights three key needs: (1) broader adoption of multimodal AI, (2) development of computationally efficient and interpretable AI models, and (3) increased use of federated learning to support privacy-preserving AI training. By synthesizing insights from various studies, this paper provides a comprehensive evaluation of AI innovations in COVID-19 research and outlines key directions for future advancements in ethical and scalable AI-driven healthcare.

关键词： Artificial intelligence COVID-19 detection Federated learning multimodal data integration Privacy-preserving AI

来源：评论

学校读者我要写书评

暂无评论

Orthogonalized Kernel Debiased Machine Learning for multimodal data Analysis

引用

JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION 2023年第543期118卷 1796-1810页

作者： Dai, Xiaowu Li, Lexin Univ Calif Berkeley Dept Econ Berkeley CA 94720 USA Univ Calif Berkeley Dept Elect Engn & Comp Sci Berkeley CA USA Univ Calif Berkeley Dept Biostat & Epidemiol Berkeley CA USA

multimodal imaging has transformed neuroscience research. While it presents unprecedented opportunities, it also imposes serious challenges. Particularly, it is difficult to combine the merits of the interpretability attributed to a simple association model with the flexibility achieved by a highly adaptive nonlinear model. In this article, we propose an orthogonalized kernel debiased machine learning approach, which is built upon the Neyman orthogonality and a form of decomposition orthogonality, for multimodal data analysis. We target the setting that naturally arises in almost all multimodal studies, where there is a primary modality of interest, plus additional auxiliary modalities. We establish the root-N-consistency and asymptotic normality of the estimated primary parameter, the semi-parametric estimation efficiency, and the asymptotic validity of the confidence band of the predicted primary modality effect. Our proposal enjoys, to a good extent, both model interpretability and model flexibility. It is also considerably different from the existing statistical methods for multimodal data integration, as well as the orthogonality-based methods for high-dimensional inferences. We demonstrate the efficacy of our method through both simulations and an application to a multimodal neuroimaging study of Alzheimer's disease. Supplementary materials for this article are available online.

关键词： Basis expansion High-dimensional inference multimodal data integration Neuroimaging analysis Neyman orthogonality Reproducing kernel Hilbert space

来源：评论

学校读者我要写书评

暂无评论

Analysis of the Effects of Cross-Attention and multimodal Inputs on Robotic Assembly Tasks 24

Analysis of the Effects of Cross-Attention and Multimodal In...

引用

24th International Conference on Control, Automation and Systems

作者： Hwang, Yuhyeon Song, Jae-Bok Korea Univ Dept Mech Engn Seoul 02841 South Korea

ISBN: (纸本)9798331517939;9788993215380

This study examines the impact of effective integration of multimodal data on the performance of robotic assembly tasks. In a multimodal system, the complementary processing of different types of data plays a crucial role in enhancing the robot's ability to solve complex problems. The experiment compared the performance of neural networks using a multi-layer perceptron (MLP) structure and a cross-attention structure for processing multimodal data, which included images, force/torque, and robot posture information. The experimental results showed that the FC-CA structure, incorporating cross-attention, exhibited improved performance in terms of success rate and assembly time. These findings suggest that the cross-attention technique holds significant potential for performance improvement in the integration and interpretation of multimodal data. This study validates the effectiveness of cross-attention for enhancing the performance of multimodal data processing tasks in robotic assembly using diverse data types.

关键词： multimodal data integration Robotic Assembly

来源：评论

学校读者我要写书评

暂无评论

Integrating multimodal Patient data into Attention-Based Graph Networks for Disease Risk Prediction 22nd

Integrating Multimodal Patient Data into Attention-Based Gra...

引用

22nd International Conference on Artificial Intelligence in Medicine (AIME)

作者： Huang, Xiayuan Yale Univ New Haven CT 06510 USA

ISBN: (纸本)9783031665349;9783031665356

In the era of digital medicine, electronic health records (EHRs) encompass a vast array of patient data from diverse sources. Harnessing this readily available information is crucial for personalized medicine and predictive healthcare. This study explores the potential of integrating multimodal data into an attention-based graph network for enhancing disease risk prediction. Leveraging the intrinsic structure of medical data, we propose a novel method called MiGAT, which fuses information from various modalities, including clinical events and sequencing data. By applying graph attention networks (GAT), our model effectively captures the intricate inter-relationships among patients, particularly emphasizing genetic similarity. This enables our model to assess and prioritize disease risks based on the genetic closeness of patients. Results indicate a substantial improvement in predictive accuracy compared to baseline models, reaffirming the potential of our multimodal integration strategy. Our findings underscore the importance of synergizing diverse EHR data sources through advanced computational models for better-informed clinical decision-making and more effective patient care.

关键词： EHR Machine Learning Disease Risk Modeling multimodal data integration Health Informatics

来源：评论

学校读者我要写书评

暂无评论

EmoStream: Real-time Emotion Prediction through Speech with Apache Flink and Kafka integration 15

EmoStream: Real-time Emotion Prediction through Speech with ...

引用

15th International Conference on Computing Communication and Networking Technologies, ICCCNT 2024

作者： Karri, Naga Bhuvana Ponamala, Ishitha Gudla, Vijayatejaswari Vemula, Swarna Raju Koneru Lakshmaiah Education Foundation Department of Computer Science and Engineering Andhra Pradesh Vaddeswaram India

ISBN: (纸本)9798350370249

Advancements in real-time emotion recognition systems have witnessed a surge in interest, fuelled by their potential applications in diverse fields such as human computer interaction, mental health monitoring, and user experience enhancement. Existing approaches often focus on singular modalities, such as speech analysis or facial expressions. However, a comprehensive understanding of human emotion necessitates the integration of multiple modalities. This research introduces a novel approach, a multimodal Real-time Emotion Prediction System (MREPS), which combines speech analysis with streaming data processing. The system begins by capturing a continuous stream of audio data, which undergoes Speech-to-Text (STT) conversion using cloud-based services like Google Cloud Speech-to-Text. Simultaneously, facial expression recognition and physiological signal analysis can be integrated for a more comprehensive emotional profile. The real-time nature of the data necessitates the use of a robust streaming architecture, leveraging technologies such as Apache Kafka for data ingestion and Apache Flink for stream processing. A crucial component of the system is the Emotion Prediction API, a microservice housing a machine learning model for emotion classification. This model is trained on diverse datasets to ensure accuracy across various languages, cultures, and individual differences. The real-time API endpoint continuously processes the multimodal input, providing dynamic emotion predictions. © 2024 IEEE.

关键词： Apache Flink Apache Kafka Cloud-based services Docker Emotion Prediction API Facial expression recognition Human-computer interaction Kubernetes Machine learning model Mental health monitoring Microservices multimodal data integration multimodal emotion prediction system Physiological signal analysis Privacy Real-time emotion recognition Security measures Speech analysis Speech-to Text Streaming architecture User feedback

来源：评论

学校读者我要写书评

暂无评论

Intricacies of single-cell multi-omics data integration

引用

TRENDS IN GENETICS 2022年第2期38卷 128-139页

作者： Rautenstrauch, Pia Vlot, Anna Hendrika Cornelia Saran, Sepideh Ohler, Uwe Max Delbruck Ctr Mol Med Berlin Inst Med Syst Biol D-10115 Berlin Germany Humboldt Univ Dept Comp Sci D-10117 Berlin Germany Humboldt Univ Dept Biol D-10117 Berlin Germany

A wealth of single-cell protocols makes it possible to characterize different molecular layers at unprecedented resolution. Integrating the resulting multi modal single-cell data to find cell-to-cell correspondences remains a challenge. We argue that data integration needs to happen at a meaningful biological level of abstraction and that it is necessary to consider the inherent discrepancies between modalities to strike a balance between biological discovery and noise removal. A survey of current methods reveals that a distinction between technical and biological origins of presumed unwanted variation between datasets is not yet commonly considered. The increasing availability of paired multimodal data will aid the development of improved methods by providing a ground truth on cell-to-cell matches.

关键词： single-cell omics multimodal data integration single-cell multi-omic assays cell type identity method development multi-omics data integration

来源：评论

学校读者我要写书评

暂无评论

Holistic Characterization of a Salmonella Typhimurium Infection Model Using Integrated Molecular Imaging

引用

JOURNAL OF THE AMERICAN SOCIETY FOR MASS SPECTROMETRY 2021年第12期32卷 2791-2802页

作者： Strittmatter, Nicole Kanvatirth, Panchali Inglese, Paolo Race, Alan M. Nilsson, Anna Dannhorn, Andreas Kudo, Hiromi Goldin, Robert D. Ling, Stephanie Wong, Edmond Seeliger, Frank Serra, Maria Paola Hoffmann, Scott Maglennon, Gareth Hamm, Gregory Atkinson, James Jones, Stewart Bunch, Josephine Andren, Per E. Takats, Zoltan Goodwin, Richard J. A. Mastroeni, Pietro Univ Cambridge Dept Vet Med Cambridge CB3 0ES England AstraZeneca R&D Biopharmaceut Discovery Sci Alderley Pk Macclesfield SK10 4TG Cheshire England Imperial Coll London Fac Med Dept Metab Digest & Reprod London SW7 2AZ England AstraZeneca Clin Pharmacol & Safety Sci R&D Cambridge CB4 0WG England Uppsala Univ Med Mass Spectrometry Imaging Dept Pharmaceut Biosci S-75124 Uppsala Sweden Uppsala Univ Sci Life Lab Spatial Mass Spectrometry S-75124 Uppsala Sweden Imperial Coll London Div Digest Dis Sect Pathol St Marys Hosp London W2 1NY England Charing Cross Hosp Dept Cellular Pathol London W6 8RF England AstraZeneca Biol Engn R&D Cambridge CB4 0WG England Univ Edinburgh BHF Ctr Cardiovasc Sci Queens Med Res Inst Edinburgh EH16 4TJ Midlothian Scotland Natl Phys Lab Natl Ctr Excellence Mass Spectrometry Imaging NiC Teddington TW11 0LW Middx England Univ Glasgow Inst Infect Immun & Inflammat Coll Med Vet & Life Sci Glasgow G12 8TA Lanark Scotland Tech Univ Munich Fac Chem D-85748 Garching Germany

A more complete and holistic view on host-microbe interactions is needed to understand the physiological and cellular barriers that affect the efficacy of drug treatments and allow the discovery and development of new therapeutics. Here, we developed a multimodal imaging approach combining histopathology with mass spectrometry imaging (MSI) and same section imaging mass cytometry (IMC) to study the effects of Salmonella Typhimurium infection in the liver of a mouse model using the S. Typhimurium strains SL3261 and SL1344. This approach enables correlation of tissue morphology and specific cell phenotypes with molecular images of tissue metabolism. IMC revealed a marked increase in immune cell markers and localization in immune aggregates in infected tissues. A correlative computational method (network analysis) was deployed to find metabolic features associated with infection and revealed metabolic clusters of acetyl carnitines, as well as phosphatidylcholine and phosphatidylethanolamine plasmalogen species, which could be associated with pro-inflammatory immune cell types. By developing an IMC marker for the detection of Salmonella LPS, we were further able to identify and characterize those cell types which contained S. Typhimurium.

关键词： mass spectrometry imaging imaging mass cytometry Salmonella infection network analysis multimodal data integration

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：