Modeling cardiac/cerebrovascular variability from wearable biometrics

从可穿戴生物识别技术中建模心脏/脑血管变异性使用小型廉价的可穿戴设备（如Apple Watch和Samsung Galaxy Watch），对心率变异性（HRV）等生物识别技术进行大规模、全人群的监测现在是可行的。智能手表能够通过手腕处的光学传感进行长期HRV监测，从而全天候连续可靠地测量生理状态。原始光学信号是包含关于许多HRV指数的信息的周期函数，包括：静息心率、活动心率、灌注指数、呼吸性窦性心律失常和昼夜节律相关的心率依赖性。尽管这些设备的消费者基础不断扩大，潜力明显（2018年售出了5000万只Apple Watch，具有经过验证的房颤检测功能），但连续光学智能手表记录中包含的信息在很大程度上仍未被利用。我们仍然缺乏对大脑/心脏解剖和功能的个体间变异性如何与一般人群的HRV相关的基本了解，特别是使用智能手表测量的长期HRV。举例来说：如果一个人在深度睡眠期间表现出不规则的HRV，这是否反映了他们大脑/心脏解剖或功能的潜在问题？ 因此，本研究的主要目标是构建一个模型，将智能手表的HRV测量与从磁共振成像（MRI）获得的心脏和脑血管测量相关联。这将通过以下方式实现：1）使用商用可穿戴设备对普通人群的样本进行长期心率变异性监测，2）在同一组受试者中获取高分辨率定量心脏/脑血管MRI测量值，以及3）建立模型来连接两个数据集。 通过建立一个能够基于智能手表衍生的HRV预测个人MRI扫描的模型，我们将朝着真正全面的大规模监测普通人群以预测疾病和死亡迈出第一步。西方世界的许多疾病是高度可预防的;据估计，50%的心血管死亡率可以通过完全消除危险因素来避免，25%的中风被认为是高度可预防的。该项目将研究智能手表衍生的HRV在多大程度上可以预测健康人群的大脑和心脏变异性，证明使用智能手表预测更昂贵的措施（如MRI）的可行性，并最终给出脑血管和心血管事件的早期预警信号。

Modeling cardiac/cerebrovascular variability from wearable biometrics Mass, population-wide monitoring of biometrics such as heartrate variability (HRV) is now feasible using small, inexpensive wearables such as Apple Watch and Samsung Galaxy Watch. Smartwatches are capable of long-term HRV monitoring through optical sensing at the wrist, yielding a continuous and reliable measure of physiological state 24/7. The raw optical signal is a periodic function containing information about many HRV indices, including: resting heart rate, active heart rate, perfusion index, respiratory sinus arrhythmia, and circadian-linked heartrate dependencies. Despite the expanding consumer base and clear potential of these devices (50 million Apple Watches sold in 2018, with validated atrial fibrillation detection), the information contained in continuous optical smartwatch recordings remains largely untapped. We still lack a basic understanding of how inter-individual variability in cerebral/cardiac anatomy and function relates to HRV across the general population, and in particular, long term HRV measured using smartwatches. For example: if an individual exhibits irregular HRV during deep sleep, does it reflect an underlying problem in their cerebral/cardiac anatomy or function? Hence, the principal objective of this research is to construct a model linking smartwatch measures of HRV to cardiac and cerebrovascular measures obtained from magnetic resonance imaging (MRI). This will be accomplished by 1) performing long-term HRV monitoring using a commercial wearable device in a sample from the general population, 2) acquiring high resolution quantitative cardiac/cerebrovascular MRI measures in the same group of subjects, and 3) building a model to link the two datasets. By building a model capable of predicting an individual's MRI scans based on smartwatch-derived HRV, we will take the first step towards a truly comprehensive, mass-scale monitoring of the general population to predict illness and death. Many disease of the western world are highly preventable; it is estimated that 50% of cardiovascular mortality can be avoided with the complete elimination of risk factors, and 25% of all strokes are considered highly preventable. This project will investigate the extent to which smartwatch-derived HRV can predict cerebral and cardiac variability across a healthy population, demonstrating the feasibility of using smartwatches to predict more expensive measures such as MRI, and eventually, give early warning signs of cerebrovascular and cardiovascular incidents.