Dodge Hiroko H, Estrin Deborah
Kevreson Professor of Neurology at University of Michigan, and Professor of Neurology at Oregon Health & Science University.
Tishman Professor of Computer Science at Cornell Tech, Cornell University.
Bridge (Wash D C). 2019 Spring;49(1):39-46.
All people are uniquely endowed at birth by genetic and environmental conditions; by the time they enter their last decades, they have a lifetime of differentiation that determines their state of health and response to new events and conditions. This cumulative differentiation creates substantial intraindividual variability in the rate of aging as well as the extent of resistance and resilience to pathological insults. Therefore, applying normative group data such as group means or median thresholds often fails to accurately identify and predict an individual's clinical state and prognosis. There are two ways to cope with this high intraindividual variability. One is to use "big data," which consists of a large number of subjects to improve the prediction algorithm. Another is to use each subject as their own universe to identify subtle changes or deviations from their premorbid stage. Rich temporal data from a single person-what we call "small data"-can be used for the individual's tailored diagnosis, disease management, and health behavior. Using such data for patient care, self-care, sustained independence, and research involves access to, processing, and interpretive use of an individual's combined data streams over time.
所有人在出生时都由基因和环境条件赋予了独特的特质；到他们进入人生的最后几十年时，他们已经历了一生的分化过程，这决定了他们的健康状况以及对新事件和新状况的反应。这种累积的分化在衰老速度以及对病理损伤的抵抗力和恢复力方面造成了个体内部的显著差异。因此，应用规范性的群体数据，如群体均值或中位数阈值，往往无法准确识别和预测个体的临床状态及预后。有两种方法可以应对这种高度的个体内部差异。一种是使用“大数据”，它由大量受试者组成，以改进预测算法。另一种是将每个受试者自身视为一个整体，以识别与其病前阶段相比的细微变化或偏差。来自单个人的丰富时间数据——我们称之为“小数据”——可用于个体的个性化诊断、疾病管理和健康行为。将这些数据用于患者护理、自我护理、持续独立生活和研究，涉及对个体随时间变化的综合数据流的获取、处理和解释性使用。
