Benchmarking Real-Time Algorithms for In-Phase Auditory Stimulation of Low Amplitude Slow Waves With Wearable EEG Devices During Sleep

Objective: In-phase stimulation of EEG slow waves (SW) during deep sleep has shown to improve cognitive function. SW enhancement is particularly desirable in subjects with low-amplitude SW such as older adults or patients suffering from neurodegeneration. However, existing algorithms to estimate the up-phase of EEG suffer from a poor phase accuracy at low amplitudes and when SW frequencies are not constant. Methods: We introduce two novel algorithms for real-time EEG phase estimation on autonomous wearable devices, a phase-locked loop (PLL) and, for the first time, a phase vocoder (PV). We compared these phase tracking algorithms with a simple amplitude threshold approach. The optimized algorithms were benchmarked for phase accuracy, the capacity to estimate phase at SW amplitudes between 20 and 60 mu V, and SW frequencies above 1 Hz on 324 home-based recordings from healthy older adults and Parkinson disease (PD) patients. Furthermore, the algorithms were implemented on a wearable device and the computational efficiency and the performance was evaluated in simulation and with a PD patient. Results: All three algorithms delivered more than 70% of the stimulation triggers during the SW up-phase. The PV showed the highest capacity on targeting low-amplitude SW and SW with frequencies above 1 Hz. The hardware testing revealed that both PV and PLL have marginal impact on microcontroller load, while the efficiency of the PV was 4% lower. Active stimulation did not influence the phase tracking. Conclusion: This work demonstrated that phase-accurate auditory stimulation can also be delivered during fully remote sleep interventions in populations with low-amplitude SW.

Automated summary

This study introduces two novel algorithms for real-time EEG phase estimation and compares them with a traditional method. The results show that both new algorithms can accurately estimate the phase of EEG signals, even at low amplitudes and non-constant frequencies. These algorithms perform better in targeting low-amplitude and higher frequency SW. Hardware testing demonstrates that these algorithms have little impact on microcontroller load.