Objective:

Sleep architecture and microstructures alter with aging. This study was to investigate the association between a sleep EEG-based brain age index (BAI), predicted brain age minus chronological age, and shift work.

Background:

Shift work is associated with numerous negative health consequences, and BAI might reflect an individual’s brain health.

Design/Methods:

We have developed a deep learning technique to estimate brain aging from EEG from sleep polysomnogram by using DenseNet to train the brain age prediction model with a six-channel sleep EEG. To allow for EEG-BAI to capture brain aging more accurately, we have taken advantage of a BIG-DATA-driven machine learning approach with 4,215 sleep EEG. Then, we enrolled 12h-shift female nurses working at one university-affiliated hospital (n=37, mean age 28.9 y, shift work duration 5.4 y). Daytime sleep study in laboratory were adjusted to the habitual sleep hour after completing the 1st 12h-night shift and eating breakfast. They were clustered based on EEG-BAI and shift working period. We used K-means clustering method to divide into two groups. The group with low BAI group was defined as the good resilience group (GR), and the group with high BAI group was defined as the poor resilience group (PR).

Results:

Of the total nurses, 26 were classified into the GR and 11 into the PR. Compared with the PR, GR started shift work at younger age (22.4 vs. 24.1 y, p<0.001), had more shift work experience (6.5 vs. 2.9 y, p<0.005), younger chronological age (27.4 vs. 29.6 y, p<0.005), and lower BAI (-0.4 vs. 1.0, p<0.001).

Conclusions:

Our results demonstrate that the EEG-BAI can be a biomarker reflecting brain health for shift workers. To adjust well to shift work, it is recommended to start shift work at a young age.