Calibrating free-living physical activity characteristics across functionally-limited populations using machine-learned accelerometer approaches

Regular physical activity is known to have positive effects on many health outcomes. In research, physical activity is often measured using activity monitors worn on the hip, wrist or ankle. These instruments determine how much physical activity a person does based on the amount of movement the person does. This is done with equations that use the data collected by the activity monitors. However, these equations are usually developed using measurements on healthy people. People who have movement disorders related to diseases such as Parkinson?s and Multiple Sclerosis, or conditions such as knee replacement, stroke, or arthritis have very different movement patterns than people without these diseases. Thus, the equations used to convert activity monitor data will not work well in people with these diseases or conditions. The purpose of this study is to develop equations to measure physical activity using activity monitors in people with movement limitations. Rather than creating specific equations for each individual disease and condition, we will perform simple tests to measure upper and lower body function, and then group people into different clusters based on these tests. Our preliminary studies show that these equations are more accurate than equations based on specific diseases or conditions.
MoreAdult
Observational
Local

Edward Melanson, PhD, FACSM
Protocol Number: 16-2706
Is this Study for You?
Not finding the right Study for you? Join ResearchMatch, a nation-wide registry connecting volunteers and researchers