Calibration Of The Biotrainer Pro Activity Monitor In Youth Using Receiver Operator Characteristic (roc) Curves

Abstract

To facilitate the application of accelerometry-based monitors for field research with children, monitors must be calibrated with age-appropriate activities and with a criterion measure that is sensitive to intermittent changes in the intensity and duration of physical activity. PURPOSE The purpose of this study was to develop calibration equations for the Biotrainer Pro that capture children's intermittent activity patterns. METHODS The participants in this study included 9 boys and 13 girls (ages 7 to 12) that were participating in a youth sport camp. Participants wore a HR monitor and a Biotrainer while completing 2 testing protocols 1) a structured protocol involving 7 intermittent movement tasks (2 min each) and 2) a free play situation (∼15 minutes) in which the child participated in a group activity. Both protocols were videotaped and coded using an adapted version of the Children's Activity Record System. Coding was done with a computerized observation tool (BEST) that allows “real time” coding of data. The outcome measure, computed as the mean observation code across each minute (1=sit, 2=stand, 3=walk, and 4=run), was temporally merged with corresponding HR and Biotrainer data. Receiver operator characteristic (ROC) curves were used to determine the cutpoint defining the threshold for moderate to vigorous activity in youth (operationalized as a code of 2.5). This approach offers considerable advantages for selecting cutpoints for accelerometry-based activity monitors since it can empirically determine the appropriate balance between sensitivity and specificity. RESULTS A two-way ANOVA yielded non significant gender and age group effects so combined data were used. Observation codes ranged from 1.0 (HR ∼ 100) to 3.99 (HR ∼ 182) for the structured calibration while the mean code for the free-play activities was 2.3 (HR ∼ 140). ROC curves yielded cutpoints of 3.0 using both the structured protocol and the free play protocol. The curves demonstrated good discrimination characteristics with sensitivities and specificity both exceeding .78. This indicates that the cutpoint can classify periods of physical activity while maintaining good discriminatory capacity to exclude non-active moments. CONCLUSIONS The results support the use of the Biotrainer Pro calibration cutpoint for youth ages 7–12.