检索结果-内蒙古大学图书馆

SSRN 2025年

作者： Ahmed, Abdulaziz Saleem, Mohammad Alzeen, Mohammed Birur, Badari Fargason, Rachel E. Burk, Bradley G. Harkins, Hannah Rose Alhassan, Ahmed Al-Garadi, Mohammed Ali Department of Health Services Administration School of Health Professions University of Alabama at Birmingham BirminghamAL United States Department of Biomedical Informatics and Data Science Heersink School of Medicine University of Alabama at Birmingham BirminghamAL35233 United States Department of Psychiatry and Behavioral Neurobiology University of Alabama at Birmingham BirminghamAL United States Department of Pharmacy University of Alabama at Birmingham BirminghamAL United States UAB Center for Psychiatric Medicine University of Alabama at Birmingham BirminghamAL United States Department of Biomedical Informatics School of Medicine Vanderbilt University Medical Center NashvilleTN United States

Background: Emergency department (ED) returns for mental health conditions represent a significant healthcare burden. Traditional machine learning models for predicting these returns often lack interpretability for clinical implementation. Methods: To evaluate whether integrating large language models (LLMs) with traditional machine learning approaches improves both the predictive accuracy and clinical interpretability of ED mental health returns models. This retrospective cohort study analyzed 42,464 ED visits for 27,904 unique mental health patients at an Academic Medical Center in the deep South of the U.S. between January 2018 and December 2022. Two outcomes were evaluated: (1) 30-day ED return prediction accuracy and (2) model interpretability through a novel LLM-enhanced explainability framework integrating SHAP (SHapley Additive exPlanations) values with contextual clinical knowledge. Findings: For chief complaint classification, Llama 3 (8B) with 10-shot learning outperformed traditional models, achieving 0.882 accuracy and 0.86 F1-score. In SDoH classification, LLM-based models achieved F1 scores between 0.67-0.96 for Alcohol, Tobacco, Substance Abuse, Exercise, and Home Environment. These results demonstrate the effectiveness of LLM-enhanced feature extraction in clinical prediction. The proposed machine learning interpretability framework, leveraging LLM, achieved 99% accuracy in translating model predictions into clinically relevant explanations. LLM-extracted features improved XGBoost's AUC from 0.74 to 0.76 and AUC-PR from 0.58 to 0.61. Interpretation: Integrating LLMs with traditional machine learning models yielded modest improvements in ED return prediction accuracy while substantially enhancing model interpretability. This approach offers a framework for translating complex predictive analytics into actionable clinical insights. © 2025, The Authors. All rights reserved.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Fast sequence to graph alignment using the graph wavefront algorithm

arXiv

引用

arXiv 2022年

作者： Zhang, Haowen Wu, Shiqi Aluru, Srinivas Li, Heng School of Computational Science and Engineering Georgia Institute of Technology United States Department of Mathematics Nanjing University China Institute for Data Engineering and Science Georgia Institute of Technology United States Department of Data Sciences Dana-Farber Cancer Institute United States Department of Biomedical Informatics Harvard Medical School United States

Motivation: A pan-genome graph represents a collection of genomes and encodes sequence variations between them. It is a powerful data structure for studying multiple similar genomes. Sequence-to-graph alignment is an essential step for the construction and the analysis of pan-genome graphs. However, existing algorithms incur runtime proportional to the product of sequence length and graph size, making them inefficient for aligning long sequences against large graphs. Results: We propose the graph wavefront alignment algorithm (Gwfa), a new method for aligning a sequence to a sequence graph. Although the worst-case time complexity of Gwfa is the same as the existing algorithms, it is designed to run faster for closely matching sequences, and its runtime in practice often increases only moderately with the edit distance of the optimal alignment. On four real datasets, Gwfa is up to four orders of magnitude faster than other exact sequence-to-graph alignment algorithms. We also propose a graph pruning heuristic on top of Gwfa, which can achieve an additional ∼10-fold speedup on large graphs. © 2022, CC BY.

关键词： Wavefronts

来源：评论

学校读者我要写书评

暂无评论

Empowering biomedical Discovery with AI Agents

arXiv

引用

arXiv 2024年

作者： Gao, Shanghua Fang, Ada Huang, Yepeng Giunchiglia, Valentina Noori, Ayush Schwarz, Jonathan Richard Ektefaie, Yasha Kondic, Jovana Zitnik, Marinka Department of Biomedical Informatics Harvard Medical School BostonMA United States Department of Chemistry and Chemical Biology Harvard University CambridgeMA United States Program in Biological and Biomedical Sciences Harvard Medical School BostonMA United States Department of Brain Sciences Imperial College London London United Kingdom Harvard College CambridgeMA United States Program in Biomedical Informatics Harvard Medical School BostonMA United States Department of Electrical Engineering and Computer Science MIT CambridgeMA United States Kempner Institute for the Study of Natural and Artificial Intelligence Harvard University MA United States Broad Institute of MIT and Harvard CambridgeMA United States Harvard Data Science Initiative CambridgeMA United States

We envision "AI scientists" as systems capable of skeptical learning and reasoning that empower biomedical research through collaborative agents that integrate AI models and biomedical tools with experimental platforms. Rather than taking humans out of the discovery process, biomedical AI agents combine human creativity and expertise with AI’s ability to analyze large datasets, navigate hypothesis spaces, and execute repetitive tasks. AI agents are poised to be proficient in various tasks, planning discovery workflows and performing self-assessment to identify and mitigate gaps in their knowledge. These agents use large language models and generative models to feature structured memory for continual learning and use machine learning tools to incorporate scientific knowledge, biological principles, and theories. AI agents can impact areas ranging from virtual cell simulation, programmable control of phenotypes, and the design of cellular circuits to developing new therapies. © 2024, CC BY.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Augmenting biomedical Named Entity Recognition with General-domain Resources

arXiv

引用

arXiv 2024年

作者： Yin, Yu Kim, Hyunjae Xiao, Xiao Wei, Chih Hsuan Kang, Jaewoo Lu, Zhiyong Xu, Hua Fang, Meng Chen, Qingyu Department of Computer Science University of Liverpool LiverpoolL69 3DR United Kingdom Department of Computer Science Korea University 145 Anam-ro Seongbuk-gu Seoul02841 Korea Republic of National Center for Biotechnology Information National Library of Medicine National Institutes of Health BethesdaMD0894 United States Department of Biomedical Informatics and Data Science School of Medicine Yale University New HavenCT06510 United States

Objective: Training a neural network-based biomedical named entity recognition1 (BioNER) model usually requires extensive and costly human annotations. While several studies have employed multi-task learning with multiple BioNER datasets to reduce human effort, this approach does not consistently yield performance improvements and may introduce label ambiguity in different biomedical corpora. We aim to tackle those challenges through transfer learning from easily accessible resources with fewer concept overlaps with biomedical datasets. Methods: We proposed GERBERA, a simple-yet-effective method that utilized general-domain NER datasets for training. We performed multi-task learning to train a pre-trained biomedical language model with both the target BioNER dataset and the general-domain dataset. Subsequently, we fine-tuned the models specifically for the BioNER dataset. Results: We systematically evaluated GERBERA on five datasets of eight entity types, collectively consisting of 81,410 instances. Despite using fewer biomedical resources, our models demonstrated superior performance compared to baseline models trained with additional BioNER datasets. Specifically, our models consistently outperformed the baseline models in six out of eight entity types, achieving an average improvement of 0.9% over the best baseline performance across eight entities. Our method was especially effective in amplifying performance on BioNER datasets characterized by limited data, with a 4.7% improvement in F1 scores on the JNLPBA-RNA dataset. Conclusion: This study introduces a new training method that leverages cost-effective general-domain NER datasets to augment BioNER models. This approach significantly improves BioNER model performance, making it a valuable asset for scenarios with scarce or costly biomedical datasets. We make data, codes, and models publicly available via https://***/qingyu-qc/bioner_gerbera. Copyright © 2024, The Authors. All rights reserved.

关键词： Multi-task learning

来源：评论

学校读者我要写书评

暂无评论

Optic nerve head assessment in Light-field fundus images - A case study

引用

ACTA OPHTHALMOLOGICA 2022年第S267期100卷

作者： Dietzel, Alexander Schramm, Stefan Blum, Maren-Christina Link, Dietmar Klee, Sascha Institute of Biomedical Engineering and Informatics Optoelectrophysiological Engineering Group Technische Universität Ilmenau Ilmenau Germany Department of General Health Studies Division Biostatistics and Data Science Karl Landsteiner University of Health Sciences Krems Austria

Purpose A suspicious optic nerve head (ONH) appearance, with abnormal cupping or increased cup-to-disc ratio, is associated to a strong risk factor for glaucoma. Typical devices to analyse the structure of the ONH are optical coherence tomography or scanning laser ophthalmoscopy. Those devices provide parameters to describe the status of an ONH like its depth, width or volume. We show that these parameters for glaucoma diagnosis can be obtained with a commercial fundus camera using an adapted light field (LF) camera providing a one-shot 3d retinal image. Methods We examined a human eye (32 years, male) 5 times during one session in compliance with the Declaration of Helsinki. The images were taken with a fundus camera (FF450, Carl Zeiss Meditec AG, Germany) combined with an industrial LF camera (RX12, Raytrix GmbH, Germany). The generic depth estimation was performed with the Raytrix LF software (RxLive 5.0.031.0) using the Depth Raycast Light algorithm. The results were exported as 3d point lists in ascii XYZ format and for further processing imported in Matlab (R2018b, MathWorks, U.S.). The imported data were pre-processed, the papilla midpoints automatically detected and the optic nerve head was horizontally sliced in 0.01 virtual depth steps starting from the base upwards until the papilla rim was reached. We compared the overall virtual depth, the virtual volume and the virtual area per virtual depth step, each as relative difference. Results All LF images showed visually a very good image quality. The depth estimations were successful and covered the complete image areas. The overall virtual depth range was 95.9% to 101.4% with mean of 100% (±2.36%). The overall virtual volume range was 97.5% to 102.2% with mean of 100% (±2.1%). The area per virtual depth range was 2.5% to 174.5% with mean of 100% (±10.9%) whereas the larger deviations appeared in the lowermost slices. Conclusions Fundus cameras with LF technique are able to provide one-shot 3d retinal images.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Heightened COVID-19 Mortality in People With Severe Mental Illness Persists After Vaccination: A Cohort Study of Greater Manchester Residents (vol 49, pg 275, 2023)

引用

SCHIZOPHRENIA BULLETIN 2023年第5期49卷 1399-1403页

作者： Hassan, Lamiece Sawyer, Chelsea Peek, Niels Lovell, Karina Carvalho, Andre F. Solmi, Marco Tilston, George Sperrin, Matthew Firth, Joseph Division of Psychology and Mental Health The University of Manchester Manchester Academic Health Science Centre Manchester M13 9PL UK Centre for Health Informatics Division of Informatics Imaging and Data Sciences The University of Manchester M13 9PL UK NIHR Greater Manchester Patient Safety Translational Research Centre The University of Manchester Manchester UK Manchester Academic Health Science Centre National Institute for Health Research Manchester Biomedical Research Centre The University of Manchester Manchester M13 9PL UK Division of Nursing Midwifery and Social Work University of Manchester Manchester Academic Health Science Centre Manchester M13 9PL UK IMPACT (Innovation in Mental and Physical Health and Clinical Treatment) Strategic Research Centre School of Medicine Barwon Health Deakin University Geelong Victoria Australia Psychiatry Department University of Ottawa Ottawa ON Canada The Ottawa Hospital University of Ottawa Ottawa ON Canada Clinical Epidemiology Program Ottawa Hospital Research Institute (OHRI) University of Ottawa Ottawa ON Canada Greater Manchester Mental Health NHS Foundation Trust Manchester Academic Health Science Centre Manchester M13 9PL UK

Background and hypothesis Previous studies show that people with severe mental illness (SMI) are at higher risk of COVID-19 mortality, however limited evidence exists regarding risk post-vaccination. We investigated COVID-19 mortality among people with schizophrenia and other SMIs before, during and after the UK vaccine roll-out. Study Design Using the Greater Manchester (GM) Care Record to access routinely collected health data linked with death records, we plotted COVID-19 mortality rates over time in GM residents with schizophrenia/psychosis, bipolar disorder (BD) and/or recurrent major depressive disorder (MDD) from February 2020 to September 2021. Multivariable logistic regression was used to compare mortality risk (risk ratios;RRs) between people with SMI (N=190,188) and age-sex matched controls (N=760,752), adjusted for sociodemographic factors, pre-existing comorbidities and vaccination *** Results Mortality risks were significantly higher among people with SMI compared with matched controls, particularly among people with schizophrenia/psychosis (RR 3.14, CI 2.66-3.71) and/or BD (RR 3.17, CI 2.15-4.67). In adjusted models, the relative risk of COVID-19 mortality decreased, though remained significantly higher than matched controls for people with schizophrenia (RR 1.53, CI 1.24-1.88) and BD (RR 2.28, CI 1.49-3.49), but not recurrent MDD (RR 0.92, CI 0.78-1.09). People with SMI continued to show higher mortality rate ratios relative to controls throughout 2021, during vaccination roll-out. Conclusions People with SMI, notably schizophrenia and bipolar disorder, were at greater risk of COVID-19 mortality compared to matched controls. Despite population vaccination efforts that have prioritised people with SMI, disparities still remain in COVID-19 mortality for people with SMI.

关键词：

来源：评论

学校读者我要写书评

暂无评论

MOVE: Integrating Multi-source Information for Predicting DTI via Cross-view Contrastive Learning

MOVE: Integrating Multi-source Information for Predicting DT...

引用

IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

作者： Yuening Qu Chengxin He Jin Yin Zhenjiang Zhao Jingyu Chen Lei Duan School of Computer Science Sichuan University Chengdu China Med-X Center for Informatics Sichuan University Chengdu China The West China Biomedical Big Data Center West China Hospital Sichuan University Chengdu China

ISBN: (数字)9781665468190

ISBN: (纸本)9781665468206

Drug-target interaction (DTI) prediction serves as the foundation of new drug findings and drug repositioning. For drugs/targets, the sequence data contains the biological structural information, while the heterogeneous network contains the biochemical functional information. These two types of information describe different aspects of drugs and targets. Due to the complexity of DTI machinery, it is necessary to learn the representation from multiple perspectives. We hereby try to design a way to leverage information from multi-source data to the maximum extent and find a strategy to fuse them. To address the above challenges, we propose a model, named MOVE (short for integrating multi-source information for predicting DTI via cross-view contrastive 1earning), for learning comprehensive representations of each drug and target from multi-source data. MOVE extracts information from the sequence view and the network view, then utilizes a fusion module with auxiliary contrastive learning to facilitate the fusion of representations. Experimental results on the benchmark dataset demonstrate that MOVE is effective in DTI prediction.

关键词： Drugs Fuses Biological system modeling Predictive models Heterogeneous networks data models Complexity theory

来源：评论

学校读者我要写书评

暂无评论

OP0269 DEVELOPMENT AND EVALUATION OF MACHINE LEARNING ALGORITHMS FOR THE PREDICTION OF OPIOID-RELATED DEATHS AMONG UK PATIENTS WITH NON-CANCER PAIN

引用

Annals of the Rheumatic Diseases 2023年 82卷 177-178页

作者： J. Benitez-Aurioles D. Jenkins C. Ramirez Medina N. Peek M. Jani The University of Manchester Division of Informatics Imaging and Data Science Manchester United Kingdom The University of Manchester NIHR Greater Manchester Patient Safety Translational Research Centre Manchester United Kingdom The University of Manchester Centre for Epidemiology Versus Arthritis Manchester United Kingdom Manchester University NHS Foundation Trust Manchester Academic Health Science Centre NIHR Manchester Biomedical Research Centre Wythenshawe United Kingdom

Background There has been a sharp rise in opioid use for non-cancer pain globally, including for rheumatic and musculoskeletal conditions. Despite increased awareness of adverse effects, they remain commonly prescribed in most countries. Clinical prediction models (CPMs) offer the possibility of assessing individual risk allowing better allocation of resources towards those at risk. Machine learning (ML) approaches can address nonlinear relationships and complex interactions between variables and are increasingly used to develop CPMs. Objectives To develop, validate, and compare the performance of three CPMs based on regression and ML, which leverage primary care data to estimate the risk of opioid-related death in patients prescribed opioids for non-cancer pain. Methods Patients ≥18 years old without prior cancer who were prescribed any opioid between 01/01/2006 and 31/12/2017 were identified in the Clinical Practice Research datalink (CPRD), representative of national patient data from UK primary care. Only new opioid users were included. Index date was date of first prescription, with censoring at withdrawal from the CPRD or after not having an opioid prescription for two years. Baseline data were extracted from each patient's records, including demographic information, comorbidities, concomitant medications, and the opioid type being prescribed. 49 candidate predictors were used to train three competing risk models: a Fine&Gray regression model with LASSO regularisation, a survival random forest (RF), and a neural network (DeepHit). The outcome was opioid-related mortality and other cause mortality the competing event, defined using a curated ICD-10 codelist. Predictive performance of the models, like area under the receiver characteristic operator curve (AUCROC) were calculated using 5-fold cross validation. Results We included a total of 1,029,681 patients, of which 1,240 experienced an opioid-related death, and 52,833 experienced a competing death. The Fine&G

关键词：

来源：评论

学校读者我要写书评

暂无评论

Neuronal sources of visually evoked potentials using selective color opponent channel stimulation

引用

ACTA OPHTHALMOLOGICA 2022年第S267期100卷

作者： Klee, Sascha Link, Dietmar Department of General Health Studies Division Biostatistics and Data Science Karl Landsteiner University of Health Sciences Krems Austria Institute for Biomedical Engineering and Informatics Department of Optoelectrophysiological Engineering Technische Universität Ilmenau Ilmenau Germany

Purpose The retinal stages of color processing have been extensively researched in recent years. In contrast, the cortical stages, especially the spatial distribution of visual processing are less understood. It is known that many cells in the visual cortex respond to color. However, what kind of computations these cells perform on their visual input, and what are their temporal and spatial properties are still under debate. In this study, we aimed to analyze the neuronal sources of visually evoked potentials using selective color opponent channel stimulations. Methods The study was conducted on 10 healthy subjects (6 male, 4 female, 1 eye each, mean age 25.5±5.1 yr). All subjects were free of eye diseases. Normal color vision was checked with Ishihara and Stilling-Velhagen plates. To modulate selective activity in the S-cone and L/M-cone driven channels (the related cells are: short-wavelength absorbing: S-cone, medium-wavelength absorbing: M-cone, long-wavelength-absorbing: L-cone) we presented two silent substitution flash sequences (total of 200 stimuli) in a full circle with a diameter of 22°. Electroencephalography was performed using a 64-channel amplifier system (Theraprax, neuroConn GmbH, Germany) in combination with Ag/AgCl ring electrodes (B10-S-200, EASYCAP GmbH, Germany). The preprocessed data were analyzed with source localization software (CURRY v4.6, Compumedics, USA). For each subject, a realistically shaped volume conductor was constructed by a boundary element model. Results We found two main components (N1, P1) in all subjects, as well as a third late component in most subjects. For these components significant latency differences (mean: N1 = 33 ms, P1 = 22 ms; p < 0.05) between both color opponent channels were found. The results showed no differences in the topography and no differences in dipole localization between both color channels. Talairach coordinates of grand averages indicated activation in area 18. Conclusions Our findings showed tha

关键词：

来源：评论

学校读者我要写书评

暂无评论

Explaining medical AI performance disparities across sites with confounder Shapley value analysis

arXiv

引用

arXiv 2021年

作者： Wu, Eric Wu, Kevin Zou, James Department of Electrical Engineering Stanford University United States Department of Biomedical Informatics Stanford University United States Department of Biomedical Data Science Stanford University United States

Medical AI algorithms can often experience degraded performance when evaluated on previously unseen sites. Addressing cross-site performance disparities is key to ensuring that AI is equitable and effective when deployed on diverse patient populations. Multi-site evaluations are key to diagnosing such disparities as they can test algorithms across a broader range of potential biases such as patient demographics, equipment types, and technical parameters. However, such tests do not explain why the model performs worse. Our framework provides a method for quantifying the marginal and cumulative effect of each type of bias on the overall performance difference when a model is evaluated on external data. We demonstrate its usefulness in a case study of a deep learning model trained to detect the presence of pneumothorax, where our framework can help explain up to 60% of the discrepancy in performance across different sites with known biases like disease comorbidities and imaging parameters. © 2021, CC BY.

关键词： Deep learning

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：