The Effect of the Philips Audio Neurofeedback System on Sleep

Abstract

Introduction This study aims to see if the PNFS has an effect on sleep parameters. Sleep deprivation has significant adverse effects (e.g., Bolge, Doan, Kannan, & Baran, 2009; Gustavsson et al., 2011). Sleep medication, although improving sleep, can have adverse effects on daytime functioning (e.g., Verster, Veldhuijzen,& Volkerts, 2004). The PNFS may present an alternative method to improve sleep quality, by means of Audio EEG neurofeedback. This is used as an relaxation technique; participants are rewarded when the EEG signal indicates relaxation. Hypotheses/research questions The research questions are whether subjectively reported sleep parameters (PSQI, TWT, NAWand SOL) improve as the PNFS is used, and whether the SMR power increased most and the beta power decreased most, when the audio neurofeedback condition was aiming to achieve this. The hypotheses are that they can be answered in the positive. Method 60 healthy participants will be included. Neurofeedback training is performed at home, in the evening, for half an hour, at least 21 days. Participants are equally divided among 3 groups. In the SMR up group, participants are rewarded with inclusion of bass-frequencies in their favourite music when they have relatively high SMR power. In the Beta down group participants receive this reward when beta power decreases. In the Placebo group the participants are given a random sample from previous recordings. Results are analysed using linear mixed effects models (Bates, M¨achler, Bolker, & Walker, 2015). Results Only the results from the first 30 participants are presented in this paper. Only the PSQI decreased significantly more in the training groups than in the Placebo group. The EEG results show that the SMR power at the start of each session increased the least in the SMR up group. The beta power at the start of each session decreased the most in the SMR up group. Each consecutive session, participants in the Beta down group do get faster at decreasing beta power. In the SMR up group, participants get faster at increasing the SMR power. For both groups there were significant differences in baseline coefficients. As a result, it is unclear whether baseline differences between groups caused the improvements, or the intervention. Conclusion Further analyses is required, before definite conclusions can be drawn. The models for the EEG power should include additional random effects, in order to control for baseline differences. Moreover, the relation between the subjective response variables and the EEG power should be assessed. The screens in the app that potentially provide visual neurofeedback should be blocked during neurofeedback training. The use of the Health Watch in Phase 2 may provide further insight in the relation between objective sleep measures and neurofeedback, additional questionnaires might include response variables that are influenced by the neurofeedback.