Validation of Smartphone-Derived Digital Phenotypes for Cognitive Assessment in Older Adults

项目摘要/摘要 目前，每10个65岁及以上的美国人中就有一个患有阿尔茨海默氏症。此外，全球范围内流行的 预计到2050年，痴呆症人数将增加两倍，从5000万增加到1.5亿以上。鉴于这一趋势，有必要 日益高效、客观和敏感的方法来表征认知、评估风险和区分 处于疾病谱不同阶段的个体。传统的神经心理评估一直是 为这些目的进行了广泛验证，但存在几个方法学缺陷，如有限 生态有效性、实践效果和繁重的测试和评分程序，这些都是人类容易 错误。另一方面，智能手机无处不在，老年人拥有智能手机的比例越来越高，而且 提供一种独特的方法来捕捉日常生活中微妙的认知和行为概况，而不是 实验室或诊所。在拟议的观察性研究中，我们的目标是探索数字表型的有效性。 作为一种新的方法来描述一组不同的老年人的认知和功能 成年人。具体地说，我们将调查被动捕获的基于智能手机的数字 特征和黄金标准的神经心理测量。我们还将探索最佳采样率，以 收集数字数据以及临床上有用的数字表型，为未来的研究提供信息。总共有90个 参与者年龄在65岁及以上，认知正常，轻度认知障碍，轻度阿尔茨海默病 将从费城地区和已完成的合格参与者池中招聘 最近的衰老研究。在为期四周的研究中，参与者将自然地使用自己的个人智能手机 当安全软件应用程序不引人注意且连续地获取未识别的原始传感器时- 基于跨域的数据，包括设备活动和使用、空间轨迹和移动性以及社交 互动。通过研究软件提供的每日调查将用于补充被动收集的 数据。参与者还将在基线访问中完成传统的神经心理学测量，以检查 构建效度。这项研究将探索数字表型是否可以提供一种有效、高效和 对认知、功能和疾病负担进行横断面追踪的自然主义方法， 随着时间的推移。如果成功，该工具可以应用于几种临床和研究环境，以产生更好的效果 评估效率、更高的诊断准确性和个性化治疗干预， 联合起来将产生巨大的成本节约和改善的健康结果。一项培训计划已经完成 设计咨询了数字表型、日常认知和功能领域的专家 老龄化人口，以及安全的计算机系统，以发展申请人在设计和 使数字表型平台用于老龄化人口，纵向数据分析用于连续 多变量数据，以及与个人数字数据相关的安全保护。

PROJECT SUMMARY/ABSTRACT At present, one in 10 Americans age 65 and older has Alzheimer's dementia. Further, the global prevalence of dementia is expected to triple from 50 million to over 150 million by 2050. Given this trend, there is a need for increasingly efficient, objective, and sensitive methods to characterize cognition, assess risk, and discriminate individuals at various stages of the disease spectrum. Traditional neuropsychological assessments have been extensively validated for these purposes, yet present several methodological drawbacks such as limited ecological validity, practice effects, and burdensome testing and scoring procedures that are prone to human error. On the other hand, smartphones are ubiquitous, are owned by older adults at increasing rates, and present a unique method to capture subtle cognitive and behavioral profiles in everyday life, outside of the laboratory or clinic. In the proposed observational study, we aim to explore the validity of a digital phenotyping protocol as a novel method for characterizing cognition and function among a heterogeneous group of older adults. Specifically, we will investigate the relations between passively captured smartphone-based digital features and gold-standard neuropsychological measures. We also will explore optimal sampling rates for collection of digital data along with clinically-useful digital phenotypes to inform future studies. A total of 90 participants age 65 and older with normal cognition, mild cognitive impairment, and mild Alzheimer's dementia will be recruited from the Philadelphia region and from a pool of eligible participants who have completed recent aging studies. Participants will use their own personal smartphones naturally during a four-week study period while a secure software application unobtrusively and continuously obtains de-identified raw sensor- based data spanning domains including device activity and usage, spatial trajectories and mobility, and social interactions. Daily surveys delivered via the study software will be used to complement the passively collected data. Participants also will complete traditional neuropsychological measures at a baseline visit to examine construct validity. This study will explore whether digital phenotyping may provide a valid, highly efficient, and naturalistic method for tracking cognition, function and disease burden both cross-sectionally and, eventually, over time. If successful, this tool can be applied in several clinical and research contexts to yield improved assessment efficiency, enhanced diagnostic accuracy, and personalized treatment interventions, which together will generate tremendous cost savings and improved health outcomes. A training plan has been designed in consultation with experts in the fields of digital phenotyping, everyday cognition and function in aging populations, and secure computer systems to develop the applicant's expertise in designing and adapting digital phenotyping platforms for use in aging populations, longitudinal data analysis for continuous multivariate data, and security protections related to personal digital data.