检索结果-内蒙古大学图书馆

Prediction of Oral Food Challenge Outcomes via Ensemble Learning

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhang, Justin Lee, Deborah Jungles, Kylie Shaltis, Diane Najarian, Kayvan Ravikumar, Rajan Sanders, Georgiana Gryak, Jonathan Department of Electrical and Computer Engineering University of Michigan Ann ArborMI United States Department of Internal Medicine University of Michigan Ann ArborMI United States Department of Pediatrics University of Michigan Ann ArborMI United States Department of Computational Medicine and Bioinformatics University of Michigan Ann ArborMI United States Michigan Institute for Data Science University of Michigan Ann ArborMI United States Department of Emergency Medicine University of Michigan Ann ArborMI United States Department of Computer Science and Engineering University of Michigan Ann ArborMI United States Max Harry Weil Institute for Critical Care Research and Innovation University of Michigan Ann ArborMI United States Mary H. Weiser Food Allergy Center University of Michigan Ann ArborMI United States Department of Computer Science Queens College City University of New York New YorkNY United States

Oral Food Challenges (OFCs) are essential to accurately diagnosing food allergy due to the limitations of existing clinical testing. However, some patients are hesitant to undergo OFCs, while those willing suffer from limited access to allergists in rural/community healthcare settings. Despite its success in predicting patient outcomes in other clinical settings, few applications of machine learning to food allergy have been developed. Thus, in this study, we seek to leverage machine learning methodologies for OFC outcome prediction. Retrospective data was gathered from 1,112 patients who collectively underwent a total of 1,284 OFCs, and consisted of clinical factors including serum-specific Immunoglobulin E (IgE), total IgE, skin prick tests (SPTs), comorbidities, sex, and age. Using these features, multiple machine learning models were constructed to predict OFC outcomes for three common allergens: peanut, egg, and milk. The best performing model for each allergen was an ensemble of random forest (egg) or Learning Using Concave and Convex Kernels (LUCCK) (peanut, milk) models, which achieved an Area under the Curve (AUC) of 0.91, 0.96, and 0.94, in predicting OFC outcomes for peanut, egg, and milk, respectively. Moreover, all such models had sensitivity and specificity values ≥ 89%. Model interpretation via SHapley Additive exPlanations (SHAP) indicates that specific IgE, along with wheal and flare values from SPTs, are highly predictive of OFC outcomes. The results of this analysis suggest that ensemble learning has the potential to predict OFC outcomes and reveal relevant clinical factors for further study. Copyright © 2022, The Authors. All rights reserved.

关键词： Machine learning

Improving Policy-Oriented Agent-Based Modeling with History Matching: A Case Study

学校读者我要写书评

暂无评论

arXiv 2024年

作者： O’Gara, David Kerr, Cliff C. Klein, Daniel J. Binois, Mickaël Garnett, Roman Hammond, Ross A. Division of Computational and Data Sciences Washington University in St. Louis St. LouisMO United States Institute for Disease Modeling Bill & Melinda Gates Foundation SeattleWA United States Acumes Team Université Côte d’Azur Inria Sophia Antipolis France Department of Computer Science McKelvey School of Engineering Washington University in St. Louis St. LouisMO United States Public Health Brown School Washington University in St. Louis St. LouisMO United States Center on Social Dynamics and Policy Brookings Institution WashingtonDC United States The Santa Fe Institute Santa FeNM United States

Advances in computing power and data availability have led to growing sophistication in mechanistic mathematical models of social dynamics. Increasingly these models are used to inform real-world policy decision-making, often with significant time sensitivity. One such modeling approach is agent-based modeling, which offers particular strengths for capturing spatial and behavioral realism, and for in-silico experiments (varying input parameters and assumptions to explore their downstream impact on key outcomes). To be useful in the real-world, these models must be able to qualitatively or quantitatively capture observed empirical phenomena, forming the starting point for subsequent experimentation. computational constraints often form a significant hurdle to timely calibration and policy analysis in high-resolution agent-based models. In this paper, we present a technical solution to address this bottleneck, substantially increasing efficiency and thus widening the range of utility for policy models. We illustrate our approach with a case study using a previously published and widely used epidemiological model. © 2024, CC BY.

关键词： Digital elevation model

Analysis of node2vec random walks on networks

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Meng, Lingqi Masuda, Naoki Department of Mathematics University at Buffalo State University of New York BuffaloNY14260-2900 United States Computational and Data-Enabled Science and Engineering Program University at Buffalo State University of New York BuffaloNY14260-5030 United States

Random walks have been proven to be useful for constructing various algorithms to gain information on networks. Algorithm node2vec employs biased random walks to realize embeddings of nodes into low-dimensional spaces, which can then be used for tasks such as multi-label classification and link prediction. The performance of the node2vec algorithm in these applications is considered to depend on properties of random walks that the algorithm uses. In the present study, we theoretically and numerically analyze random walks used by the node2vec. Those random walks are second-order Markov chains. We exploit the mapping of its transition rule to a transition probability matrix among directed edges to analyze the stationary probability, relaxation times in terms of the spectral gap of the transition probability matrix, and coalescence time. In particular, we show that node2vec random walk accelerates diffusion when walkers are designed to avoid both back-tracking and visiting a neighbor of the previously visited node but do not avoid them completely. Copyright © 2020, The Authors. All rights reserved.

关键词： Relaxation time

Meta-analysis of computational methods for breast cancer classification

学校读者我要写书评

暂无评论

International Journal of Intelligent Information and database Systems 2020年第1期13卷 89-111页

作者： Pham, Tri-Cong Luong, Chi-Mai Doucet, Antoine Hoang, Van-Dung Tran, Diem-Phuc Le, Duc-Hau School of Computer Science and Engineering Thu-yloi University 175 Tay Son Dong Da Hanoi Viet Nam Department of Informatics and Communication Technology University of Science and Technology of Hanoi Hanoi Viet Nam Vietnam Academy of Science and Technology 18 Hoang Quoc Viet Cau Giay Hanoi Viet Nam Institute of Information Technology Vietnam Academy of Science and Technology 18 Hoang Quoc Viet Cau Giay Hanoi Viet Nam Laboratory L3i University of La Rochelle France QuangBinh University Dong Hoi Quang Binh Viet Nam DuyTan University Da Nang Viet Nam Department of Computational Biomedicine Vingroup Big Data Institute No 7 Bang Lang 1 Street Viet Hung Ward Hanoi Viet Nam

Millions of women are suffering from breast cancer pressing burden on their shoulders and the global economy. Meanwhile, general treatment methods are applied without considering personalised health and genetic features. Artificial intelligence appears to be a robust method for breast cancer sub-typing. Most of researches have been implemented on binary classification with limited number of data samples. Multi-classification is much more difficult especially on large number of samples. The study aims to use machine learning to find better ways to subtype breast cancer as well as find new disease causative genes which help facilitate more personalised treatment with limited side effect in the future. This study compares the accuracy of three classification methods in combination with eight feature selection methods on a dataset of 2,682 samples. The study shows that the highest accuracy was 83.96% with the SVM-C005 classifier and percentile feature selection (800 genes). Additionally, our method can predict causative disease genes of breast cancer with four of them known to be associated with breast cancer and 29 promising ones with supporting evidence from the literature. This shows the effectiveness of our research. Copyright © 2020 Inderscience Enterprises Ltd.

关键词： Gene expression

Identifying Reasons for Contraceptive Switching from Real-World data Using Large Language Models

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Miao, Brenda Y. Williams, Christopher Y.K. Chinedu-Eneh, Ebenezer Zack, Travis Alsentzer, Emily Butte, Atul J. Chen, Irene Y. Bakar Computational Health Sciences Institute University of California San Francisco San FranciscoCA United States Department of Medicine University of California San Francisco San FranciscoCA United States Helen Diller Family Comprehensive Cancer Center University of California San Francisco San FranciscoCA United States Division of General Internal Medicine Brigham and Women's Hospital BostonMA United States Harvard Medical School BostonMA United States Center for Data-driven Insights and Innovation University of California Office of the President OaklandCA United States Computational Precision Health University of California Berkeley University of California San Francisco BerkeleyCA United States Electrical Engineering and Computer Science University of California Berkeley BerkeleyCA United States Berkeley AI Research University of California Berkeley BerkeleyCA United States

Background: Understanding why patients switch contraceptives is of significant interest but these factors are often only captured in unstructured clinical notes and can be difficult to extract. We evaluate the zero-shot abilities of a recently developed large language model, GPT-4, to identify reasons for switching between classes of contraceptives from clinical notes. Methods: Clinical notes associated with contraceptive switches, defined as a difference in prescribed contraceptive modality between consecutive encounters, were selected from the UCSF Information Commons dataset. GPT-4 (HIPAA-compliant Microsoft Azure API) was used to extract contraceptive class started, stopped, and reasons for switching from these clinical notes. We evaluated six prompts on a subset of human annotated notes, and extracted reasons for contraceptive switching from the full dataset using the best prompt based on microF1 scores. BERTopic topic modeling was used to identify key reasons for contraceptive switching and enrichment of topics was assessed in patient subgroups stratified by race and ethnicity. Results: We extracted 1,964 contraceptive switches from 1,515 patients. Younger patients and patients with self-reported race/ethnicities listed as "Black/African American" and "Latinx" had significantly higher rates of switching. GPT-4 demonstrated extraction of contraceptives started and stopped had MicroF1 scores of 0.849 and 0.881, respectively. Accuracy of GPT-4 extracted reasons for switching was 91.4% and 2.2% contained hallucinations. Key reasons for switching included patient preference, adverse events, and insurance. "Weight gain/mood change" and "insurance coverage" topics were enriched in patients self-reporting as being "Black or African American", "Latinx", or "Multi-Race/Ethnicity". Conclusion: We demonstrate that GPT-4 can accurately extract reasons for contraceptive switching and identify certain reasons that are disproportionately found in specific demographic populati

关键词： Extraction

EXPLAINABLE ARTIFICIAL INTELLIGENCE (XAI) 2.0: A MANIFESTO OF OPEN CHALLENGES AND INTERDISCIPLINARY RESEARCH DIRECTIONS

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Longo, Luca Brcic, Mario Cabitza, Federico Choi, Jaesik Confalonieri, Roberto Ser, Javier Del Guidotti, Riccardo Hayashi, Yoichi Herrera, Francisco Holzinger, Andreas Jiang, Richard Khosravi, Hassan Lecue, Freddy Malgieri, Gianclaudio Páez, Andrés Samek, Wojciech Schneider, Johannes Speith, Timo Stumpf, Simone The Artificial Intelligence and Cognitive Load Research Lab Technological University Dublin Ireland University of Zagreb Faculty of Electrical Engineering and Computing Croatia University of Milano-Bicocca Milan Italy IRCCS Ospedale Galeazzi Sant’Ambrogio Milan Italy Kim Jaechul Graduate School of AI Korea Advanced Institute of Science & Technology Korea Republic of INEEJI Corporation Korea Republic of Department of Mathematics University of Padua Italy Derio Spain Bilbao Spain University of Pisa Pisa Italy Department of Computer Science Meiji University Tokyo Japan Department of Computer Science and Artificial Intelligence DaSCI Andalusian Institute in Data Science & Computational Intelligence University of Granada Granada Spain Human-Centered AI Lab University of Natural Resources and Life Sciences Vienna Austria School of Computing and Communications Lancaster University United Kingdom The University of Queensland Brisbane Australia Sophia Antipolis France eLaw Center for Law and Digital Technologies Leiden University Netherlands Department of Philosophy Universidad de los Andes Bogotá Colombia Center for Research & Formation in Artificial Intelligence Universidad de los Andes Bogotá Colombia Technical University of Berlin Berlin Germany Fraunhofer Heinrich Hertz Institute Berlin Germany Berlin Germany Department of Information Systems and Computer Science University of Liechtenstein Liechtenstein Liechtenstein Department of Philosophy University of Bayreuth Bayreuth Germany Center for Perspicuous Computing Saarland University Saarbrücken Germany School of Computing Science University of Glasgow United Kingdom

As systems based on opaque Artificial Intelligence (AI) continue to flourish in diverse real-world applications, understanding these black box models has become paramount. In response, Explainable AI (XAI) has emerged as a field of research with practical and ethical benefits across various domains. This paper not only highlights the advancements in XAI and its application in real-world scenarios but also addresses the ongoing challenges within XAI, emphasizing the need for broader perspectives and collaborative efforts. We bring together experts from diverse fields to identify open problems, striving to synchronize research agendas and accelerate XAI in practical applications. By fostering collaborative discussion and interdisciplinary cooperation, we aim to propel XAI forward, contributing to its continued success. Our goal is to put forward a comprehensive proposal for advancing XAI. To achieve this goal, we present a manifesto of 27 open problems categorized into nine categories. These challenges encapsulate the complexities and nuances of XAI and offer a road map for future research. For each problem, we provide promising research directions in the hope of harnessing the collective intelligence of interested stakeholders. © 2023, CC BY-SA.

关键词： Ethical technology

One-dimensional deep low-rank and sparse network for accelerated MRI

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Wang, Zi Qian, Chen Guo, Di Sun, Hongwei Li, Rushuai Zhao, Bo Qu, Xiaobo Department of Electronic Science Biomedical Intelligent Cloud R&D Center Fujian Provincial Key Laboratory of Plasma and Magnetic Resonance National Institute for Data Science in Health and Medicine Xiamen University China School of Computer and Information Engineering Xiamen University of Technology Xiamen China United Imaging Research Institute of Intelligent Imaging Beijing China Department of Nuclear Medicine Nanjing First Hospital Nanjing Medical University Nanjing China Department of Biomedical Engineering Oden Institute for Computational Engineering and Sciences University of Texas at Austin Austin United States

Deep learning has shown astonishing performance in accelerated magnetic resonance imaging (MRI). Most state-of-the-art deep learning reconstructions adopt the powerful convolutional neural network and perform 2D convolution since many magnetic resonance images or their corresponding k-space are in 2D. In this work, we present a new approach that explores the 1D convolution, making the deep network much easier to be trained and generalized. We further integrate the 1D convolution into the proposed deep network, named as One-dimensional Deep Low-rank and Sparse network (ODLS), which unrolls the iteration procedure of a low-rank and sparse reconstruction model. Extensive results on in vivo knee and brain datasets demonstrate that, the proposed ODLS is very suitable for the case of limited training subjects and provides improved reconstruction performance than state-of-the-art methods both visually and quantitatively. Additionally, ODLS also shows nice robustness to different undersampling scenarios and some mismatches between the training and test data. In summary, our work demonstrates that the 1D deep learning scheme is memory-efficient and robust in fast MRI. Copyright © 2021, The Authors. All rights reserved.

关键词： Magnetic resonance imaging

Generative models of simultaneously heavy-tailed distributions of interevent times on nodes and edges

学校读者我要写书评

暂无评论

Physical Review E 2020年第5期102卷 052303-052303页

作者： Elohim Fonseca dos Reis Aming Li Naoki Masuda Department of Mathematics State University of New York at Buffalo Buffalo New York 14260 USA Department of Zoology University of Oxford Oxford OX1 3PS United Kingdom Department of Biochemistry University of Oxford Oxford OX1 3QU United Kingdom Computational and Data-Enabled Science and Engineering Program State University of New York at Buffalo Buffalo New York 14260 USA Faculty of Science and Engineering Waseda University 169-8555 Tokyo Japan

Intervals between discrete events representing human activities, as well as other types of events, often obey heavy-tailed distributions, and their impacts on collective dynamics on networks such as contagion processes have been intensively studied. The literature supports that such heavy-tailed distributions are present for interevent times associated with both individual nodes and individual edges in networks. However, the simultaneous presence of heavy-tailed distributions of interevent times for nodes and edges is a nontrivial phenomenon, and its origin has been elusive. In the present study, we propose a generative model and its variants to explain this phenomenon. We assume that each node independently transits between a high-activity and low-activity state according to a continuous-time two-state Markov process and that, for the main model, events on an edge occur at a high rate if and only if both end nodes of the edge are in the high-activity state. In other words, two nodes interact frequently only when both nodes prefer to interact with others. The model produces distributions of interevent times for both individual nodes and edges that resemble heavy-tailed distributions across some scales. It also produces positive correlation in consecutive interevent times, which is another stylized observation for empirical data of human activity. We expect that our modeling framework provides a useful benchmark for investigating dynamics on temporal networks driven by non-Poissonian event sequences.

关键词： Network structure Social networks Stochastic networks Network Models

Emergency Department Decision Support using Clinical Pseudo-notes

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Lee, Simon A. Jain, Sujay Chen, Alex Ono, Kyoka Rudas, Akos Fang, Jennifer Chiang, Jeffrey N. Department of Computational Medicine University of California Los AngelesCA90095 United States Department of Electrical and Computer Engineering University of California at Los Angeles Los AngelesCA90095 United States Department of Statistics and Data Science University of California Los Angeles United States Department of Natural Sciences International Christian University Mitaka Tokyo Japan LA Health Services Enterprise Clinical Informatics Los AngelesCA United States Harbor-UCLA Medical Center Department of Emergency Medicine TorranceCA United States University of California Los Angeles Department of Emergency Medicine Los AngelesCA United States Department of Neurosurgery University of California Los AngelesCA90095 United States

In this work, we introduce the Multiple Embedding Model for EHR (MEME), an approach that serializes multimodal EHR tabular data into text using "pseudo-notes", mimicking clinical text generation. This conversion not only preserves better representations of categorical data and learns contexts but also enables the effective employment of pretrained foundation models for rich feature representation. To address potential issues with context length, our framework encodes embeddings for each EHR modality separately. We demonstrate the effectiveness of MEME by applying it to several decision support tasks within the Emergency Department across multiple hospital systems. Our findings indicate that MEME outperforms traditional machine learning, EHR-specific foundation models, and general LLMs, highlighting its potential as a general and extendible EHR representation strategy. © 2024, CC BY.

关键词： Embeddings