AI-ADRD: Accelerating interventions of AD/ADRD via Machine learning methods

AI-ADRD：通过机器学习方法加速 AD/ADRD 干预

• 批准号: 10682237
• 负责人: Jiang Bian
• 金额: $ 234.89万
• 依托单位: UNIVERSITY OF PENNSYLVANIA
• 依托单位国家: 美国
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-05-01 至 2026-04-30
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10682237
• 关键词: Acceleration; Active Learning; Alzheimer' s Disease; Alzheimer' s disease related dementia; Behavioral; Characteristics; Clinical; Clinical Data; Cognition; Complex; Computing Methodologies; Consumption; Data; Degenerative Disorder; Dementia; Detection; Diabetes Mellitus; Dimensions; Drug Exposure; Drug Modelings; Drug usage; Electronic Health Record; Genetic; Goals; Heterogeneity; Hyperlipidemia; Hypertension; Intervention; Investigation; Investments; Machine Learning; Manuals; Measurement; Medicine; Methods; Modeling; Neurobehavioral Manifestations; Neurodegenerative Disorders; Outcome; Patient-Focused Outcomes; Patients; Pharmaceutical Preparations; Phase; Phenotype; Physical Education; Physical activity; Prevention; Procedures; Process; Reproducibility; Research; Risk Factors; Sampling; Signal Transduction; Source; Testing; Time; United States; Work; clinical data warehouse; comorbidity; computable phenotypes; design; drug candidate; drug repurposing; drug testing; effective therapy; high dimensionality; learning strategy; machine learning method; mild cognitive impairment; multiple data sources; novel; novel therapeutics; older patient; phenotyping algorithm; response; social health determinants; statistical learning; success; supervised learning; treatment effect

PROJECT SUMMARY Alzheimer's Disease and Alzheimer's Disease Related Dementias (AD/ADRD) are neurodegenerative diseases characterized by progressive loss of cognition and other neurobehavioral symptoms. In United States, there are approximately 5.8 million patients live with AD/ADRD, and 97% of these patients are older than 65. In the last decade, real-world data (RWD), including electronic health records (EHR) data and claims data, are becoming increasingly valuable for drug repurposing of AD. In this proposal, we plan to develop novel high dimensional semi-supervised learning and active learning methods, as well as associated high dimensional hypothesis testing procedures, to accelerate interventions of AD/ADRD, using drug repurposing of AD/ADRD as a motivating use case. We will address several key challenges in real-world data including the high dimensionality of the risk factors, concomitant medication use, and complex and heterogeneous progression trajectories of AD/ADRD. The success of this project will lead to novel machine learning and statistical learning methods for facilitating drug repurposing for AD/ADRD based on real-world data.

项目总结