A novel actigraphy data analysis tool and its application to identifying the optimal threshold value in three subject...

Most actigraphy devices use different analysis methods and a non-standardized threshold value to estimate sleep/wake status and identify rest intervals. To address limitations of these approaches, a new algorithm was developed that makes no assumptions about sleep/wake status, objectively selects an optimal threshold for different populations, and provides mathematical endpoints to more fully describe the activity patterns of subjects. The optimal threshold (cts min(-1)) is defined as the value that maximizes the duration of the rest period while minimizing the inclusion of epochs from the active period. This value is identified as the beginning of a plateau region of a rest duration versus threshold value graph. Application of this new algorithm to data from 56 healthy adults, 6 healthy children, and 14 children with autism spectrum disorder (ASD) showed that the three groups had different optimal threshold values (35, 40, and 45 cts min(-1) for adults, children and ASD respectively). The rest periods of healthy children was longer than that of adults (8.5  ±  0.5 versus 6.3  ±  0.9 h, p  <  0.001). Healthy children also had less activity during the rest periods than adults (10.5  ±  1.8 versus 15.1  ±  11.8 cts min(-1)) and ASD children (12.0  ±  2.2 cts min(-1)) but these differences were not statistically significant. However, the distributions of their activity values during rest periods as measured by skewness and kurtosis were significantly greater than that of healthy adults (skewness: 7.3  ±  0.9 versus 6.2  ±  0.9, p  <  0.01, kurtosis: 83.3  ±  16.5 versus 52.8  ±  14.4, p  <  0.001) and of ASD children (skewness: 6.4  ±  0.6. p  <  0.05, kurtosis: 57.7  ±  12.8, p  <  0.001). These findings are consistent with more restful sleep patterns which would have mostly low levels of activity with few large values. The new analysis tool may be helpful in standardizing actigraphy data analyses while providing new insights into activity patterns.