Calibrating a novel multi-sensor physical activity measurement system

Abstract

Advancing the field of physical activity (PA) monitoring requires the development of innovative multi-sensor measurement systems that are feasible in the free-living environment. The use of novel analytical techniques to combine and process these multiple sensor signals is equally important. This paper, describes a novel multi-sensor ‘Integrated PA Measurement System’ (IMS), the lab-based methodology used to calibrate the IMS, techniques used to predict multiple variables from the sensor signals, and proposes design changes to improve the feasibility of deploying the IMS in the free-living environment. The IMS consists of hip and wrist acceleration sensors, two piezoelectric respiration sensors on the torso, and an ultraviolet radiation sensor to obtain contextual information (indoors vs. outdoors) of PA. During lab-based calibration of the IMS, data were collected on participants performing a PA routine consisting of seven different ambulatory and free-living activities while wearing a portable metabolic unit (criterion measure) and the IMS. Data analyses on the first 50 adult participants are presented. These analyses were used to determine if the IMS can be used to predict the variables of interest. Finally, physical modifications for the IMS that could enhance feasibility of free-living use are proposed and refinement of the prediction techniques is discussed.

Objective Assessment of Physical Activity

“…In addition, the thorough training and quality control procedures should have minimized coding errors. The sample sizes of the study cohorts were modest, although comparable to most studies of this sort (11,13,17,25,30,33). The total minutes and days of data collected, however, were much greater than previous studies.…”

“…Our algorithms performed similarly to algorithms developed using other machine learning or statistical techniques (11,13,17,25,30,33). Most previous studies have been conducted in laboratory settings (13), or with some observed data in outdoor locations (30).…”

Guide to the Assessment of Physical Activity: Clinical and Research Applications

“…[125][126][127]213 Several groups [128][129][130]159,214,215 have investigated how to extract and use more of the accelerometer signal using machinelearning algorithms to process data. These analyses provide detailed information about overall physical activity behavior, including time spent in different intensities of physical activity and activity type.…”

Computational Methods for Estimating Energy Expenditure in Human Physical Activities

“…The SVM converts the original features which usually are in a lower dimensional space to a higher dimensional space by using kernel functions such as a Gaussian radial basis function. Various studies have successfully applied ANN (52, 57) and SVM (33, 43, 53) for estimating PAEE, and promising results have been reported. For example, by applying an ANN model to signals from a biaxial accelerometer worn at hip, Rothney et al .…”