Skip to contents

Nap Detection

Source: vignettes/NapDetection.Rmd
NapDetection.Rmd

The nap detection functionality was recently introduced and has so far only been evaluated in data from pre-schoolers for hip- or wrist-worn accelerometers. These findings can hopefully be shared via journal publication in the upcoming year. Further, we are actively exploring how to best summarise the detect naps in the GGIR output. The description below reflects the current situation.

Algorithm description

The detection of both sleep and naps starts with the identification of sustained inactivity bouts as described in Chapter 8. Next, we continue with the identification of the main sleep period time window in the day as discussed in Chapter 9 and Chapter 10. The sustained inactivity bouts that do not overlap with the detected sleep period time window are detected as nap if both of the following conditions are met:

  1. Duration in minutes falls in range specified by parameter possible_nap_dur. For pre-schoolers we used c(30, 240).
  2. Timing in the day falls in range specified by parameter possible_nap_window, use c(0, 24) for full day. For pre-schoolers we used c(6, 18).

Note that default settings are possible_nap_dur=NULL and possible_nap_window=NULL meaning that nap detection will not be performed.

Time spent in naps is stored as column “dur_day_nap_min” in the GGIR part 5 csv report where it is treated as one of several mutually exclusive behavioural categories. Meaning, time spent in naps never count towards time spent in inactivity (sedentary behaviour) and visa versa.

Disclaimer

  1. Make sure that detected sleep period time (SPT) window is plausible. Without plausible SPT detection, nap detection is expected to struggle. This is particularly important for scenario where the accelerometer is not worn consistently or with highly irregular sleeping patterns.

  2. Consider using GGIR’s visualreport functionality to visually inspect the classifications before trusting the numeric output in the csv reports.

  3. Keep in mind that GGIR’s nonwear detection is sensitive to non-wear lasting at least 60 minutes. Shorter episodes of non-wear are likely to be missed and my be detected as nap.