11/11/2016

Post date: Nov 15, 2016 8:49:18 PM

Title: Statistics and Applications for Wearable Devices

Speaker: Jiawei Bai, Department of Biostatistics, JHU

[Abstract]

Wearable sensors, such as accelerometers, can now provide objective and continuous measurements for human physical activity at the sub-second level on thousands of subjects. Such devices have been widely deployed in both observational and clinical studies thanks to their rich data and low cost. In this talk I will present a big picture of accelerometry research and describe a few key research topics in this field. First I will discuss a new movement recognition approach based on “movelets”, which directly uses raw accelerometry data. I will then introduce several different metrics of physical activity, which summarize the raw data at the second or minute level. At last I will discuss a two-stage model based on the summarized minute-by-minute level data. The model describes the changes in the circadian rhythm of physical activity with respect to factors such as age, gender and body mass index.