Children diagnosed with autism spectrum disorders (ASD) often suffer from sleep disorders such as insomnia, which generates long sleep latency and fragmented sleep. Sleep disorders reduce children’s concentration for learning and contribute to increased stress for them and their families. Polysomnography (PSG) is a gold standard to evaluate and diagnose sleep patterns, but the sensors tend to be uncomfortable and expensive, and may interfere with sleep.
Actigraphy is a non-invasive method to evaluate daytime and sleep activity with a wrist device. In addition, we have developed a wireless non-invasive sensor to measure electrodermal activity (EDA) to observe sympathetic nervous activity. Combining actigraphy and EDA can provide details of children’s sleep and can be comfortably used for low-cost sleep monitoring at home..
Objectives:
We aimed to evaluate sleep patterns in children with ASD using both PSG and a wearable sensor that enables comfortable measurement of EDA through skin conductance, skin temperature, and actigraphy on the wrist.
Methods:
Six children diagnosed with ASD (ages 3-8) participated in overnight measurement in a sleep lab. One group (N=3) were good sleepers, who took melatonin before sleep and the other group (N=3) were poor sleepers. We examined skin conductance, actigraphy, and skin temperature during sleep from the inside left and right wrists (N=5, only right wrist for N=1) and compared the behavior of these signals to PSG. We obtained thirty-second epochs of labeled sleep stages (Wake, REM, Stage1, 2 and 3).
The data was analyzed as follows:
1. Pre-processing: zero-crossing and Cole's function were applied to the accelerometer data to discriminate between sleep and wake. EDA data was low-pass filtered (0.4 Hz).
2. We compared the amplitude of left and right EDA in sleep stages.
3. We analyzed “Storm” regions with high-frequency EDA, more than 6 peaks/min. We counted the number of storms per night as well as the number of peaks per storm. We also calculated areas, durations of storms and onset intervals between storms. We compared these storm characteristics to sleep stages.
Results:
Four out of five subjects showed that EDA on the left wrist was higher than that on the right wrist. Most EDA storm patterns occurred during stage 2 and 3 (slow-wave) sleep. Larger amplitude storms occurred earlier in the evening. The poor sleepers had shortened latency of the first storm (it came earlier for poor sleepers).
Conclusions:
We measured continuous EDA, actigraphy and skin temperature on children diagnosed with ASD with a comfortable wearable sensor and evaluated the relationship between EDA characteristics, laterality, and sleep stages from simultaneously recorded PSG. On most children, EDA on the left wrist was higher than EDA on the right wrist. Moreover, EDA showed characteristic high-frequency storms that occurred during stage 2 and 3 (slow-wave sleep) with larger areas under the curve earlier in the evening. The group of poor sleepers showed shorter latency of the first storms than the group of good sleepers. The comfortable wearable sensor showed new sleep characteristics on children diagnosed with ASD that could be measured easily at home.