Identifying molecular traits associated with extreme human longevity using an AI based integrative approach

使用基于人工智能的综合方法识别与人类极端长寿相关的分子特征

• 批准号: 10745015
• 负责人: Daniel Spencer Evans
• 金额: $ 23.81万
• 依托单位: UNIVERSITY OF CALIFORNIA, SAN FRANCISCO
• 依托单位国家: 美国
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-09-01 至 2025-05-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10745015
• 关键词: Acceleration; Aging; Artificial Intelligence; Biological Assay; Biological Factors; Biological Markers; Biological Process; Biological Testing; Biology; Biology of Aging; Cell Culture Techniques; Centenarian; Chronic; Clinical; Clinical Research; Cohort Studies; Collection; Complex; Computational Biology; Data; Data Set; Development; Disease; Epidemiology; Family; Foundations; Framingham Heart Study; Funding; Genomics; Human; Intervention; Investments; Knowledge; Life Expectancy; Longevity; Machine Learning; Mendelian randomization; Modeling; Molecular; Multiomic Data; Mus; Neural Network Simulation; Outcome; Pathway interactions; Performance; Phenotype; Phylogenetic Analysis; Physiology; Predictive Factor; Process; Prognostic Marker; Proteins; Research; Research Design; Testing; Time; age related; aging population; artificial intelligence method; biomarker identification; biomarker validation; cohort; data harmonization; data integration; data management; deep neural network; design; drug development; experimental study; follow-up; genetic epidemiology; healthy aging; high dimensionality; human data; molecular marker; multiple omics; neural network architecture; novel; novel marker; outcome prediction; predictive marker; risk mitigation; trait

PROJECT SUMMARY/ABSTRACT Life expectancy is increasing, and consequently, the burden of chronic age-related disease is also increasing. Interventions and treatments that target the fundamental biological process of human aging have the potential to mitigate risk of multiple diseases faced by our aging population. To develop interventions targeting the aging process, one must identify predictive factors and biomarkers associated with the aging clinical endpoints. By definition, the development of aging-based conditions and diseases takes time and requires a great deal of follow-up time. To accelerate research in human aging, biomarkers of human aging and prognostic biomarkers of healthy human aging are desperately needed. Without reliable biomarkers, early-stage drug development is severely limited. In this application, we propose a framework to identify biomarkers of healthy human aging using advanced Artificial Intelligence (AI) methods applied to a wide range of deeply phenotyped studies that collected data from humans and non-humans. We have assembled a team with deep expertise in clinical research of aging, genetic epidemiology, biology of aging, and AI. To identify biomarkers of aging through the integrative analysis of omic data with AI, we propose the following specific aims: Aim 1 (R21, first stage). Assemble datasets from the Framingham Heart Study (FHS) and the Longevity Consortium (LC) with multiple omics to test AI methods and to identify biomarkers associated with human aging. Aim 2 (R21, first stage). Test biologically informed AI Deep Neural Network (DNN) models with FHS and LC data to integrate omic data, predict outcomes, and identify predictive omic features. Aim 3 (R33, second stage). Apply models from public data onto exceptional longevity (EL) data. Aim 4 (R33, second stage). Establishing causal relationship between biomarkers and longevity phenotypes through Mendelian Randomization (MR) analysis and cell culture experiments.

项目摘要/摘要 预期寿命正在增加，因此，慢性年龄相关疾病的负担也在增加。 针对人类衰老基本生物学过程的干预措施和治疗具有潜力 减轻人口老龄化的多种疾病的风险。制定针对衰老的干预措施 过程，必须识别与衰老临床终点相关的预测因素和生物标志物。经过 定义，基于衰老的疾病和疾病的发展需要时间，需要大量 后续时间。为了加速人类衰老的研究，人类衰老和预后生物标志物的生物标志物 迫切需要健康的人类衰老。没有可靠的生物标志物，早期药物开发是 受到严格限制。在此应用中，我们提出了一个框架来识别健康人类衰老的生物标志物 使用高级人工智能（AI）方法应用于广泛的深层表型研究 收集了人类和非人类的数据。我们已经组建了一个具有深厚专业知识的团队 衰老，遗传流行病学，衰老生物学和AI的研究。通过识别衰老的生物标志物 对OMIC数据的整合分析与AI，我们提出以下特定目的：AIM 1（R21，第一阶段）。 来自Framingham心脏研究（FHS）和寿命联盟（LC）的数据集和多个 OMICS测试AI方法并识别与人衰老相关的生物标志物。目标2（R21，第一阶段）。 测试具有FHS和LC数据的生物学通知的AI深神经网络（DNN）模型以整合OMIC 数据，预测结果并识别预测性的OMIC特征。目标3（R33，第二阶段）。应用模型 公共数据到特殊的寿命（EL）数据。目标4（R33，第二阶段）。建立因果关系 通过孟德尔随机分析（MR）分析和细胞之间的生物标志物与寿命表型之间 培养实验。