Identification of Mild Cognitive Impairment using Machine Learning from Language and Behavior Markers

使用机器学习从语言和行为标记识别轻度认知障碍

• 批准号: 10709094
• 负责人: HIROKO Hayama DODGE
• 金额: $ 33.03万
• 依托单位: MICHIGAN STATE UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-04-15 至 2024-03-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10709094
• 关键词: Administrative Supplement; Affect; Algorithms; Alzheimer' s Disease; Alzheimer' s disease patient; American; Amyloid beta-Protein; Behavior; Behavior monitoring; Biological Markers; Brain imaging; Cause of Death; Cells; Clinical; Clinical Markers; Clinical Trials; Cognition; Cognitive; Cohort Studies; Data; Data Sources; Databases; Dementia; Detection; Digital biomarker; Disease Progression; Early Diagnosis; Early Intervention; Early identification; Effectiveness; Elderly; Failure; Funding; Goals; Heart Diseases; Home; Image; Individual; Language; Language Development; Lead; Learning; Machine Learning; Magnetic Resonance Imaging; Malignant Neoplasms; Measurement; Measures; Memory Loss; Modality; Modeling; Neuropsychological Tests; Outpatients; Pathologic; Patient Self-Report; Patients; Pattern; Performance; Persons; Proteomics; Reporting; Residential Facilities; Risk; Sample Size; Scientist; Signal Transduction; Structure; Test Result; Training; United States; Work; brain cell; cognitive change; cohort; computer framework; cost effective; effective therapy; imaging biomarker; improved; in vivo; insight; learning algorithm; machine learning algorithm; machine learning model; mild cognitive impairment; multimodality; novel; parent project; predictive modeling; response; screening; sensor; success; tau Proteins; transfer learning; user-friendly

Project Summary Recent estimates indicate that Alzheimer’s disease (AD) may rank as the third leading cause of death for older people in the United States, just behind heart disease and cancer. While scientists know that AD involves a progressive brain cell failure, the reason why cells fail is still not clear. To understand the progression of the disease, one of the keys is to investigate the cognitive changes in patients with mild cognitive impairment (MCI). Even though biomarkers such as imaging and clinical functions are found to be outstanding in differentiating AD patients from those with normal cognition (NC), studies suggest that their discriminative power in early-stage MCI are rather limited. Detecting signals which distinguish subjects with MCI from those with NC is challenging due to the low sensitivity and high variability of current clinical measures such as annually assessed neuropsychological test results and self-reported functional measurements. Moreover, even though in-vivo biomarkers such as beta-amyloid and tau can be used as indicators of pathological progression towards AD, the screening of biomarkers are prohibitively expensive to be widely used among pre-symptomatic individuals in the outpatient setting. We hypothesize that progressive cognitive impact from MCI has elicited detectable changes in the way people talk and behave, which can be sensed by inexpensive and accessible sensors and leveraged by machine learning (ML) algorithms to build predictive models for quantifying the risk of MCI. Our preliminary results on a small cohort indicated that there are significant differences between MCI and NC subjects during a semi-structured conversation, and ML algorithms can use such differences for differentiating MCI and NC with promising performance. Our preliminary results in behavior monitoring also suggest highly predictive performance using temporal patterns of behavior signals. In the parent project, we are building upon our initial success and conduct comprehensive studies on language and behavior markers in larger-scale cohorts to build high-performance and interpretable ML models for screening MCI. This supplement builds on our current work on digital biomarkers and will focus on further refining the prediction capability of digital biomarkers. Recently, the availability of MRI data from I-CONECT study has provided Unanticipated Opportunity for us to dramatically improve the quality of digital biomarkers. To achieve this goal, in Aim S1 we propose to develop a data-driven algorithms framework that uses high-quality imaging information as auxiliary information to increase the predictive performance of language markers; in Aim S2 we propose to develop a computational framework to use public language databases to improve the quality of language markers. This supplement, if funded, will significant predictive performance improvements of digital biomarkers and therefore improve the predictive power of early detection of MCI.

项目概要 最近的估计表明，阿尔茨海默病（AD）可能成为第三大死因 对于美国老年人来说，仅次于心脏病和癌症。虽然科学家们知道 AD涉及进行性脑细胞衰竭，但细胞衰竭的原因仍不清楚。要了解 疾病的进展，关键之一是调查轻度患者的认知变化 认知障碍（MCI）。尽管发现了影像学和临床功能等生物标志物 研究表明，在区分 AD 患者与认知正常 (NC) 患者方面表现出色 他们在早期 MCI 中的区分能力相当有限。检测区分的信号 由于敏感性低且变异性高，MCI 受试者与 NC 受试者相比具有挑战性。 当前的临床措施，例如每年评估的神经心理学测试结果和自我报告 功能测量。此外，尽管 β-淀粉样蛋白和 tau 蛋白等体内生物标志物可以 作为AD病理进展的指标，生物标志物的筛选是 在门诊环境中的症状前个体中广泛使用的费用过高。 我们假设 MCI 对认知的渐进影响已经引起了可检测到的变化 人们的说话和行为可以通过廉价且易于使用的传感器来感知并利用 通过机器学习 (ML) 算法构建预测模型来量化 MCI 风险。我们的 一个小队列的初步结果表明，MCI 和 MCI 之间存在显着差异。 半结构化对话期间的 NC 主题，ML 算法可以利用这种差异 凭借良好的性能区分 MCI 和 NC。我们在行为监测方面的初步结果 还建议使用行为信号的时间模式来实现高度预测性能。在父级中 项目中，我们正在初步成功的基础上，对语言和 大规模群体中的行为标记，以构建高性能且可解释的机器学习模型 筛查 MCI。本补充材料建立在我们当前在数字生物标记方面的工作的基础上，并将重点关注 进一步完善数字生物标志物的预测能力。最近，MRI 数据的可用性 I-CONECT 研究为我们大幅提高质量提供了意想不到的机会 数字生物标志物。为了实现这一目标，在 Aim S1 中我们建议开发一种数据驱动的算法 使用高质量成像信息作为辅助信息来增加预测的框架 语言标记的表现；在 Aim S2 中，我们建议开发一个计算框架来使用 公共语言数据库，以提高语言标记的质量。如果获得资助，该补助将 数字生物标志物的预测性能显着提高，从而提高 MCI 早期检测的预测能力。

项目成果

Corticosteroids for infectious critical illness: A multicenter target trial emulation stratified by predicted organ dysfunction trajectory.

皮质类固醇治疗传染性危重疾病：按预测的器官功能障碍轨迹分层的多中心目标试验模拟。

• DOI: 10.1101/2024.03.07.24303926
• 发表时间: 2024
• 期刊: medRxiv : the preprint server for health sciences
• 作者: Rajendran,Suraj;Xu,Zhenxing;Pan,Weishen;Zang,Chengxi;Siempos,Ilias;Torres,Lisa;Xu,Jie;Bian,Jiang;Schenck,EdwardJ;Wang,Fei
• 通讯作者: Wang,Fei

Data-Free Knowledge Distillation for Heterogeneous Federated Learning

• 发表时间: 2021-05
• 期刊: Proceedings of machine learning research
• 作者: Zhuangdi Zhu;Junyuan Hong;Jiayu Zhou

Trap and Replace: Defending Backdoor Attacks by Trapping Them into an Easy-to-Replace Subnetwork

• DOI: 10.48550/arxiv.2210.06428
• 发表时间: 2022-10
• 期刊: Advances in neural information processing systems
• 作者: Haotao Wang;Junyuan Hong;Aston Zhang;Jiayu Zhou;Zhangyang Wang

Generalizability of a Machine Learning Model for Improving Utilization of Parathyroid Hormone-Related Peptide Testing across Multiple Clinical Centers.

机器学习模型的通用性，可提高多个临床中心甲状旁腺激素相关肽测试的利用率。

• DOI: 10.1093/clinchem/hvad141
• 发表时间: 2023
• 期刊: Clinical chemistry
• 影响因子: 9.3
• 作者: Yang,HeS;Pan,Weishen;Wang,Yingheng;Zaydman,MarkA;Spies,NicholasC;Zhao,Zhen;Guise,TheresaA;Meng,QingH;Wang,Fei
• 通讯作者: Wang,Fei

Building the Model.

• DOI: 10.5858/arpa.2021-0635-ra
• 发表时间: 2023-07-01
• 期刊: Archives of pathology & laboratory medicine
• 影响因子: 4.6
• 作者: Yang HS;Rhoads DD;Sepulveda J;Zang C;Chadburn A;Wang F
• 通讯作者: Wang F