mHealth for Heart Failure: Predictive Models of Readmission Risk and Self-care Using Consumer Activity Trackers

点击翻译按钮获取中文摘要

PROJECT SUMMARY/ABSTRACT Heart failure (HF) is a debilitating disease that affects over five million people in the United States. Occurrence of, morbidity related to, and hospitalization due to HF have serious financial implications. In 2012, HF had a direct cost of over $30.7 billion annually, the majority of which was due to direct medical costs. By 2030, HF total direct costs are predicted to reach $69.7 billion, an increase of 127%. Increases in costs will be driven by an increase in the aging population, making prevention of HF and care efficiency imperative. Fifty percent of readmissions due to HF are preventable, with lack of adherence to prescribed self-care as the driving factor. Results of telemedicine intervention studies to support adherence to self-care and improve HF outcomes are inconclusive. Past telemedicine interventions for HF have utilized an array of methods including: wireless sensors, telephone services, websites, and home visits from nurses. Structured telephone support has shown in some cases to reduce hospitalization, improve clinical outcomes, and reduce all-cause mortality in HF patients. However, patient participation in telemedicine interventions varies widely. This variation is due in part to the high treatment burden placed upon patients in such home monitoring interventions, which require them to engage in novel behaviors, including using new unfamiliar hardware and spending time meeting with home health nurses. The goals of this R01 are to: 1) demonstrate that patients are adherent to a home monitoring regimen when using minimally-invasive monitoring technologies, including wrist-worn consumer activity trackers; 2) combine the minimally-invasive home monitoring regimen with predictive algorithms to forecast hospital readmission; 3) develop models using electronic health record (EHR) data and a baseline survey to predict levels of adherence to the home monitoring regimen; and 4) explore the pragmatic feasibility of using a mobile app for communicating with patients in prospective pilot study. Towards these goals, we will recruit 500 HF patients to participate in a minimally-invasive home monitoring regimen. We will measure levels of adherence to the regimen, and use collected sensor data and known readmission events to create a novel hidden semi-Markov model that continuously predicts readmission risk. Predicting a patient’s level of adherence will be performed with EHR data and a baseline survey. Finally, we will develop a mobile application that will allow patients to monitor their progress and receive adherence notifications and short surveys in a pilot study of 50 patients. The work outlined in this proposal will produce a set of foundational tools for performing home monitoring of HF patients. We will discover EHR phenotypes and mobile sensor biomarkers that are predictive of readmission and adherence, which will enable a future randomized trial that precisely targets computational patient profiles with tailored incentives based on behavioral economics to reduce hospital readmission.