UniHPF : Universal Healthcare Predictive Framework with Zero Domain Knowledge
KAIST(韩国科学技术院);NAVER AI Lab(NAVER AI实验室);Asan Medical Center, University of Ulsan College of Medicine(蔚山大学医学院峨山医学中心)
Kyunghoon Hur, Jungwoo Oh, Junu Kim, Jiyoun Kim, Min Jae Lee, Eunbyeol Cho, Seong-Eun Moon, Young-Hak Kim, Edward Choi
备注:
The original paper is published on Journal of Biomedical and Health Informatics(JBHI) 2023, https://ieeexplore.ieee.org/document/10298642. Extended Abstract presented at Machine Learning for Health (ML4H) symposium 2022, November 28th, 2022, New Orleans, United States, 19 pages(main paper 6 pages). arXiv admin note: substantial text overlap with arXiv:2207.09858
本文提出 UniHPF 通用医疗预测框架,在无需医学领域知识和极少预处理的条件下,构建可处理不同 EHR 系统异构数据的大规模预测模型,并通过实验验证其有效性。
摘要:
尽管电子健康记录(EHR)数据丰富,但其异质性限制了医疗数据在构建预测模型中的利用。为应对这一挑战,我们提出了通用医疗预测框架(UniHPF),该框架无需医学领域知识,并且对多个预测任务只需极少的预处理。实验结果表明,UniHPF 能够构建大规模 EHR 模型,处理来自不同 EHR 系统的任意形式的医疗数据。我们相信,我们的发现可为进一步研究 EHR 的多源学习提供有益的见解。
英文摘要:
Despite the abundance of Electronic Healthcare Records (EHR), its heterogeneity restricts the utilization of medical data in building predictive models. To address this challenge, we propose Universal Healthcare Predictive Framework (UniHPF), which requires no medical domain knowledge and minimal pre-processing for multiple prediction tasks. Experimental results demonstrate that UniHPF is capable of building large-scale EHR models that can process any form of medical data from distinct EHR systems. We believe that our findings can provide helpful insights for further research on the multi-source learning of EHRs.
RETAIN: An Interpretable Predictive Model for Healthcare using Reverse Time Attention Mechanism
Georgia Institute of Technology(佐治亚理工学院);Sutter Health(萨特健康)
准确性和可解释性是成功的预测模型的两个主要特征。通常情况下,必须选择支持复杂的黑盒模型,如递归神经网络(RNN)的准确性,而不是准确性较低但更可解释的传统模型,如逻辑回归。这种权衡在准确性和可解释性都很重要的医学中提出了挑战。我们通过开发用于电子健康记录(EHR)数据的REverse Time AttentIoN模型(RETAIN)来应对这一挑战。RETAIN在保持临床可解释性的同时实现了高准确性,并且基于两级神经注意力模型,该模型检测有影响力的既往访视和这些访视中的重要临床变量(例如关键诊断)。RETAIN通过以相反的时间顺序关注EHR数据来模仿医生的实践,以便最近的临床访问可能会受到更高的关注。RETAIN在一个大型健康系统EHR数据集上进行了测试,在8年的时间里,263K名患者完成了1400万次访问,并证明了与RNN等最先进方法相媲美的预测准确性和计算可扩展性,以及与传统模型相媲美的可解释性。
GenHPF: General Healthcare Predictive Framework with Multi-task Multi-source Learning
Korea Advanced Institute of Science and Technology(韩国科学技术院);Naver AI Lab(Naver AI实验室);Asan Medical Center, University of Ulsan College of Medicine(蔚山大学医学院峨山医学中心);Philips(飞利浦公司)
Massachusetts Institute of Technology(麻省理工学院);Vector Institute(向量研究所);University of Toronto(多伦多大学);New York University(纽约大学);New York University Grossman School of Medicine(纽约大学格罗斯曼医学院)
Predictive Modeling of ICU Healthcare-Associated Infections from Imbalanced Data. Using Ensembles and a Clustering-Based Undersampling Approach
Faculty of Nursing and Physiotherapy, University of Salamanca(萨拉曼卡大学护理与物理治疗学院);University Hospital of Salamanca(萨拉曼卡大学医院);Department of Computing and Automation, University of Salamanca(萨拉曼卡大学计算与自动化系);Department of Statistics, University of Salamanca(萨拉曼卡大学统计系)
Exploring Accurate and Transparent Domain Adaptation in Predictive Healthcare via Concept-Grounded Orthogonal Inference
Department of Computer Science, Stevens Institute of Technology, Hoboken, NJ, United States(斯蒂文斯理工学院计算机科学系);UMass Chan Medical School, University of Massachusetts Amherst, Amherst, MA, United States(马萨诸塞大学阿默斯特分校UMass Chan医学学校)
Who Became Financially Vulnerable After COVID-19? A Population-Level Machine Learning Analysis Using MEPS Data
Hood College(胡德学院);Georgetown University(乔治敦大学);University of Benin(贝宁大学);NIST University(NIST大学);AI CoLab: MedStar–Georgetown Collaborative Center for Artificial Intelligence in Healthcare Research and Education(人工智能合作实验室:梅斯达星 - 乔治敦医疗保健研究与教育人工智能协作中心);MedStar Health Research Institute(梅斯达星健康研究所);Albany State University(奥尔巴尼州立大学)
AI-based modular warning machine for risk identification in proximity healthcare
Department of Electrical, Computer and Biomedical Engineering, Università di Pavia, Pavia, Italy(电气、计算机与生物医学工程系,普瓦亚大学,普瓦亚,意大利);Neuroscience and Human Genetics Department, Meyer Children's Hospital IRCCS, Firenze, Italy(神经科学与人类遗传学系,梅耶儿童医院IRCCS,佛罗伦萨,意大利)
A Generative Framework for Predictive Modeling of Multiple Chronic Conditions Using Graph Variational Autoencoder and Bandit-Optimized Graph Neural Network
Department of Mechanical Engineering, The University of Texas at San Antonio(德克萨斯大学圣安东尼奥分校机械工程系);Department of Mechanical Engineering, Southern Methodist University(南方 Methodist 大学机械工程系);Engineering Technology Department, Sam Houston State University(萨姆霍茨顿州立大学工程技术系);UTHealth Houston, The University of Texas Health Science Center at Houston(德克萨斯大学健康科学中心休斯顿分校)
Integrated multimodal artificial intelligence framework for healthcare applications
Abdul Latif Jameel Clinic for Machine Learning in Health, MIT(麻省理工学院阿卜杜勒·拉蒂夫·贾米勒健康机器学习诊所);Wyss Institute for Biologically Inspired Engineering, Harvard University(哈佛大学怀斯生物启发工程研究所);Operations Research Center, Massachusetts Institute of Technology (MIT)(麻省理工学院运筹学研究中心);Sloan School of Management, MIT(麻省理工学院斯隆管理学院)
Sequential Diagnosis Prediction with Transformer and Ontological Representation
Australian Artificial Intelligence Institute, FEIT, University of Technology Sydney(悉尼科技大学澳大利亚人工智能研究所);School of Information Technology and Electrical Engineering, The University of Queensland(昆士兰大学信息技术与电气工程学院)
A Causal Framework for Mitigating Data Shifts in Healthcare
Causality in Healthcare AI Hub (CHAI)(医疗人工智能AI hub(CHAI));The University of Edinburgh(爱丁堡大学);University of Manchester(曼彻斯特大学);Athena Research Center(阿塔尼亚研究中心);NKU Athens(雅典NKU);King’s College London(伦敦国王学院);Imperial College London(伦敦帝国学院);University of Exeter(埃克塞特大学);University College London(伦敦大学学院)
A machine learning framework for interpretable predictions in patient pathways: The case of predicting ICU admission for patients with symptoms of sepsis
Technische Hochschule Nürnberg Georg Simon Ohm(纽伦堡乔治·西蒙·欧姆应用技术大学);Friedrich-Alexander-Universität Erlangen-Nürnberg(埃尔朗根-纽伦堡弗里德里希-亚历山大大学);University of Regensburg(雷根斯堡大学);Leipzig University(莱比锡大学)
Debias-CLR: A Contrastive Learning Based Debiasing Method for Algorithmic Fairness in Healthcare Applications
Wright State University(莱特州立大学);Cincinnati Children’s Hospital Medical Center(辛辛那提儿童医院医疗中心);Kairos Research LLC(Kairos Research有限责任公司);University of Illinois Chicago(伊利诺伊大学芝加哥分校)