Analyze: Machine Learning for Mobility Data

点击翻译按钮获取中文摘要

Limited mobility due to conditions like osteoarthritis (OA), cerebral palsy, and Parkinson’s disease affects millions of individuals, at enormous personal and societal cost. Rehabilitation can dramatically improve mobility and function, but current rehabilitation practice requires in-person guidance by a skilled clinician, increasing expense and limiting access. Mobile sensing technologies are now ubiquitous and have the potential to measure patient function and guide treatment outside the clinic, but they currently fail to capture the characteristics of motion required to accurately monitor function and customize treatment. Millions of low-cost mobile sensors are generating terabytes of data that could be analyzed in combination with other data, such as images, clinical records, and video, to enable studies of unprecedented scale, but machine learning models for analyzing these large-scale, heterogeneous, time-varying data are lacking. To address these challenges, we will establish a Biomedical Technology Resource Center —The Mobilize Center. Through the leadership of an experienced scientific team, we will create and disseminate innovative tools to quantify movement biomechanics with mobile sensors. Specifically, we will: 1. Push the bounds of what we can measure via wearable sensors using models that compute muscle and joint forces and metabolic cost of locomotion. These models, based on biomechanical and machine learning models, will be disseminated via our newly created OpenSense software, which will be used by thousands of researchers to gain new insights into patient biomechanics using mobile sensors. 2. Meet the need for tools that analyze data about movement dynamics and develop machine learning models to analyze and generate insights from unstructured, high-dimensional data, including time- series (e.g., from mobile sensors), images (e.g., MRI), and video (e.g., smartphone video of a patient’s gait). 3. Provide tools needed to intervene in the real-world. We will develop algorithms to accurately quantify kinematics outside the lab for long durations using data from inertial measurement units (IMUs). We will also build behavioral models to adapt and personalize goal setting, drawing on movement records from 6 million individuals, as well as health goals and exercise for 1.7 million people. Through intensive interactions with our Collaborative Projects, we will focus on improving rehabilitation outcomes for individuals with limited mobility due to osteoarthritis, obesity, Parkinson’s disease, and cerebral palsy. The Center’s tools and services will enable researchers to revolutionize how we diagnose, monitor, and treat mobility disorders, providing tools needed to deliver precision rehabilitation at low cost and on a massive scale in the future.