Development of novel polysomnography-based digital biomarkers to predict Alzheimer’s disease and Parkinson’s disease in real world settings

项目总结/摘要 在神经退行性疾病的管理中最大的未满足的挑战之一是早期诊断 阿尔茨海默病和相关痴呆症（ADRD）和帕金森病（PD）。鉴于他们的高 ADRD和PD的患病率高，前驱期长，缺乏疾病改善治疗， 至关重要。在人工智能（AI）方法的最新进展的推动下， 通过开发新的数据驱动的ADRD和PD筛选生物标志物，使用多模式， 多维、实时多导睡眠图（PSG）睡眠信号。尽管越来越多的证据表明 睡眠和ADRD/PD之间的双向关系，关于多模式PSG的实用性知之甚少 睡眠信号[例如，脑电图（EEG）用于大脑，心电图（ECG）用于心脏， 肌肉肌电图（EMG）和呼吸流量和呼吸努力]，用于识别未来的ADRD PD案件。作为一个多学科团队，拥有强大的初步数据和丰富的研究经验， 睡眠和神经退行性变的关系，我们在解决这一差距方面处于独特的地位。本提案的目的是 使用数据驱动的人工智能方法，为患者生成具有成本效益且用户友好的基于PSG的数字生物标志物。 在临床和家庭环境中预测ADRD和PD。我们的假设是PSG睡眠信号可能是 用于开发预测算法，在临床诊断前数年识别ADRD和PD， 预测算法可以从临床推广到社区环境。我们面临着前所未有的机遇 为了利用来自三个NIH支持的多中心纵向队列的数据：一个多样化的临床睡眠队列， 完整的AI睡眠报告（CAISR）研究，包括超过70，000名年龄在50岁及以上的受试者，其中15岁 的后续行动，和两个社区为基础的队列，男性骨质疏松性骨折（MrOS）睡眠研究和 骨质疏松性骨折（SOF）研究，超过3500名社区居住老年人随访13年 年使用最先进的人工智能模型，我们将追求两个具体目标：1）发现PSG生物标志物， 确定临床环境中ADRD和PD的当前和未来诊断; 2）验证性能， 在社区环境中使用家庭PSG检测ADRD和PD的PSG生物标志物的普遍性。 这将是第一项创建具有成本效益的非侵入性PSG筛选生物标志物的研究，用于识别 现实世界中的ADRD和PD。这为进一步研究无创PSG是否能作为一种有效的治疗手段奠定了基础 生物标志物可预测ADRD/PD发病机制，并可与其他创新生物标志物整合 用于改善ADRD和PD表型的表征。通过确定具有最佳的PSG模式， 预测价值，这项工作将直接告知新的用户友好型设备的长期发展 在家庭环境中监测睡眠生物标志物和ADRD/PD风险。这提供了一个变革性的公共卫生 影响，因为通过每日睡眠监测筛查ADRD和PD是可扩展的，并将识别高风险 早期诊断和早期干预。

PROJECT SUMMARY/ ABSTRACT One of the greatest unmet challenges in the management of neurodegenerative diseases is the early diagnosis of Alzheimer's disease and related dementias (ADRD) and Parkinson's disease (PD). Given their high prevalence, long prodromal period, and lack of disease-modifying therapies, early detection of ADRD and PD is of critical importance. Facilitated by recent advances in artificial intelligence (AI) methods, this proposal will break new ground by developing novel data-driven screening biomarkers for ADRD and PD, using multimodal, multidimensional, real-time polysomnography (PSG) sleep signals. Despite the growing evidence that suggests a bi-directional relationship between sleep and ADRD/PD, little is known about the utility of the multimodal PSG sleep signals [e.g., electroencephalogram (EEG) for the brain, electrocardiogram (ECG) for the heart, electromyogram (EMG) for the muscle, and respiratory flow and effort for breathing] for identifying future ADRD and PD cases. As a multidisciplinary team with strong preliminary data and extensive experiences in research of sleep and neurodegeneration, we are uniquely positioned to address this gap. The goal of this proposal is to use data-driven AI approaches to generate cost-effective and user-friendly PSG-based digital biomarkers for the prediction of ADRD and PD in clinical and at-home settings. Our hypothesis is that PSG sleep signals could be used to develop prediction algorithms that identify ADRD and PD, years before clinical diagnoses, and that the prediction algorithms can generalize from clinical to community settings. We have an unprecedented opportunity to leverage data from three NIH-supported multicenter longitudinal cohorts: a diverse clinical sleep cohort, the Complete AI Sleep Report (CAISR) study, consisting of over 70K subjects aged 50 years and older with 15 years of follow-up, and two community-based cohorts, the Osteoporotic Fractures in Men (MrOS) Sleep Study and the Study of Osteoporotic Fractures (SOF), with over 3500 community-dwelling older adults followed for up to 13 years. Using state-of-the-art AI models, we will pursue two specific aims: 1) discover PSG biomarkers that identify current and future diagnoses of ADRD and PD in clinical settings; and 2) validate the performance and generalizability of the PSG biomarkers for detecting ADRD and PD using in-home PSG in community settings. This will be the first study to create cost-effective, non-invasive PSG-based screening biomarkers for identifying ADRD and PD in real-world settings. This will set the foundation for further studying whether non-invasive PSG biomarkers are predictive of ADRD/PD pathogenesis and may be integrated with other innovative biomarkers for improved characterization of ADRD and PD phenotypes. By identifying specific PSG modalities with the best predictive value, this work will directly inform the development of new user-friendly devices for long-term monitoring of sleep biomarkers and ADRD/PD risk in home settings. This offers a transformative public health impact, as screening ADRD and PD through daily sleep monitoring is scalable and will identify high-risk individuals for early diagnosis and early intervention.