Upper-limb paresis is the most common impairment following a stroke affecting 75% of stroke survivors, which can be more prominent in one of the two limbs. Most recovery of functional impairments occurs within the first few weeks after stroke and plateaus thereafter. Unfortunately, even after patients reach a stable phase of recovery, their functional level of the stroke-affected limb may decline. Therefore, it is clinically important to maintain the regained functional level beyond the first couple weeks of spontaneous recovery by continuing to practice the use of the affected limb during daily living. Wearable technologies have emerged as a low-cost, objective tool to monitor the performance of the upper limbs during activities of daily living (ADLs). However, to date, there exists no study that has investigated the effectiveness of a mobile-health (mHealth) system aiming to enable high-dosage motor performance in chronic stroke survivors in the real-world setting. Specifically, the optimal configuration of the goal setting, feedback mechanism, and ways to share data among the stakeholders (patients and clinicians) remains unknown. This proposal aims to develop and validate an mHealth technology that aims to encourage affected limb use during the performance of ADLs in chronic stroke survivors. To accomplish this goal, we will employ the unique finger-worn ring sensor (accelerometer), developed by our academic-industry partnership, that can capture both gross-arm and fine-hand use of the limbs that are essential in the performance of ADLs. We will study important aspects of making positive behavior changes to encourage the affected limb use by fully leveraging the computational insights drawn from sensor data combined with clinical insights from providers.
To accomplish this research goal, Aim 1 will focus on the development of an mHealth platform, composed of body-networked sensors and cloud-based systems, to monitor the real-world use of the limbs in chronic stroke survivors.
In Aim 2, we will develop machine-learning-based algorithms to extract clinically meaningful information regarding real-world upper limb use from sensor data.
Aim 3 will investigate the optimal design of our mHealth system such as individual tailoring of the goal, design of the feedback, medium and timing to deliver feedback, and ways to share data among the stakeholders (patients and clinicians) via human-centered design approaches. Finally, in Aim 4, we will validate the short-term (8 weeks) effectiveness of the mHealth system in improving the use of the affected limb through a field deployment study. We believe that outcomes of this project will open a new door leading to previously unexplored datasets and understanding of patient-technology interactions to promote positive behavior changes to enable a high dosage of physical and occupational therapy, which can form the basis of a wide range of future investigations of hemiparesis rehabilitation and personalized disease management.