Estimating Trajectory of Recovery in Cardiac Rehabilitation using Mobile Health Technology and Personalized Machine Learning

项目摘要/摘要 我们的工作目标是利用移动医疗技术来开发机器学习模型 恢复的纵向轨迹，就像心脏康复所需要的那样。调查人员使用移动电话 卫生技术量化恢复措施的轨迹，个性化对运动能力的理解 和心脏功能。以运动为基础的心脏康复计划降低心血管死亡风险，并 通过增加运动能力，以这种纵向方式改善患者的预后 峰值V02在护理过程中的改进。这些计划最近已扩展到包括心脏 射血分数降低(HFrEF)的失败患者。尽管死亡率和再入院人数有所下降， 参与和坚持心脏康复方案仍然是一项挑战，特别是在服务不足的情况下。 社区，因为计划的可获得性、距离和对计划的交通访问有限，其 运营时间过长，以及缺乏多样性和以性别为主的方案。基于家庭的程序使用 智能手机已被证明能够提高忠诚度，并取得类似的结果。而基于家庭的程序 还改善了静息心率、收缩压和体力活动水平 代谢当量任务和峰值V02在研究结束时，用户表示希望有 个性化教育和治疗。基于家庭的系统仍然不能实现实时交互、反馈 并通过缺乏反馈和自我报告的必要性来监控基于中心的康复所起的作用 运动值。需要一个系统来量化运动能力的测量，这可以导致恢复， 在整个治疗过程中都是动态的。这项提议开发了一个不引人注目的系统，具有新的移动 健康技术传感器和训练分析模型，允许个性化量化康复 HFrEF患者的轨迹，可以监控患者在治疗期间的依从性和措施的改善 在休息的时候也要锻炼身体。该系统调查了12周心脏手术后的改善情况。 康复研究和设计康复轨迹以了解V02峰值和运动的改善 通过测量心率和血压测量的改善也测量HFrEF患者的容量 在休息的时候。这使得随着时间的推移，可以更好地量化复苏的其他措施进行调查 在HFrEF患者中，可用于中心康复或家庭康复。这可以提供 对定义HFrEF患者康复的指标进行了显著增强，对指标的估计为 难以收集和评估。

Project Summary/Abstract The objective of our work leverages mobile health technology to develop machine learning models for longitudinal trajectories of recovery like those needed in cardiac rehabilitation. The investigation uses mobile health technology to quantify trajectories of recovery measures, personalizing understanding of exercise capacity and cardiac function. Exercise-based cardiac rehabilitation programs reduce cardiovascular mortality risks and improve patient outcomes in such longitudinal fashion, through increased exercise capacity as measured by peak V02 improvements over the course of care. These programs have recently been extended to include heart failure with reduced ejection fraction (HFrEF) patients. Despite the reduction in mortality and readmissions, participation and adherence in cardiac rehabilitation programs remains a challenge, especially in underserved communities because of limited program availability, the distance and transportation access to a program, its hours of operation, as well as a lack of diversity and gender-dominated programs. Home-based programs using smartphones have shown to increase adherence and achieve similar outcomes. While home-based programs also improved resting heart rate, systolic blood pressure, and levels of physical activity achieved through metabolic equivalent of tasks and peak V02 at the end of the study, users expressed a desire to have individualized education and treatment. Home-based systems still do not achieve real-time interaction, feedback, and monitoring that center-based rehabilitation does through a lack of feedback and necessity of self-reported exertion values. A system is needed that quantify measures of exercise capacity, which can lead to recovery, dynamically throughout the course of treatment. This proposal develops an unobtrusive system, with new mobile health technology sensors, and trains analytic models that allow for personalized quantification of rehabilitation trajectories in HFrEF patients, which can monitor patient adherence and improvement in measures during exercise as well as while at rest. This system investigates the improvement over the course of a 12-week cardiac rehabilitation study and designs trajectories of recovery to understand improvements in peak V02 and exercise capacity in HFrEF patients by also measuring improvements of measurements of heart rate and blood pressure while at rest. This allows for an investigation of additional measures, over time, that may better quantify recovery in HFrEF patients that can be used for center-based rehabilitation or home-based rehabilitation. This can provide a significant enhancement of metrics that define recovery for HFrEF patients with estimations to metrics that are difficult to collect and evaluate.