Validity of activity measurement using a smart phone

Abstract

Since smart phones are equipped with built-in accelerometers they can be used for self-monitoring of physical activity which is an important health behavior and predictor of functioning, especially in older adults. The objective of this study is to investigate the validity of a smart phone-based activity monitoring application in adults aged below and above 65 years old. Ten adults aged below 65 years (mean of 33, standard deviation 13.7) and ten adults aged 65 years or older (mean 76, standard deviation 5.5) were asked to monitor their daily physical activity with a smart phone and an ActiGraph GT3X for 7 consecutive days. Spearman correlations between the counts per minute of the two devices were calculated for adults aged below and above 65 years separately. For both devices, each monitored minute was classified into four categories of activity intensity based on the counts per minute: sedentary, light, moderate, and high activity intensity. Association and agreement between the two devices was analyzed for separately for the two groups using Pearson's Correlations and paired t-tests. Data from 10 adults aged below 65 years and 10 adults aged above 65 years could be included in the analyses. Spearman correlations in adults aged below 65 years varied between 0.62 and 0.89 and correlations in adults aged above 65 years varied between 0.73 and 0.98. Pearson's correlations between the two devices for total number of minutes spent in different activity intensity categories per day per participant were high in both groups (range 0.79–0.98). Paired t-tests revealed that the smart phone underestimates the number of sedentary minutes per day in participants aged below and above 65 years with 6.45% and 6.78% respectively.