Implementation of mobile EEG for resting-state and visual evoked potentials in young children in rural Ethiopia

Abstract

Children living in low- and middle-income countries (LMIC) are at disproportionately higher risk of neurodevelopmental delays due to exposure to adverse biological and environmental hazards. In infancy, global developmental assessments, such as the Bayley Scales, are insensitive, do not strongly correlate with later cognitive outcomes, and require adaptation for different populations and cultural contexts. Electroencephalography (EEG) objectively measures electrical brain activity and may provide early neural markers predictive of long-term cognitive outcomes. The visual evoked potential (VEP) interrogates the efficiency of visual cortical processing and reflects neural processing speed. Mobile EEG enables the assessment of neural processing in settings where such technologies were historically inaccessible. This paper describes the experiences and lessons learned from implementing mobile EEG and VEP in rural Amhara, Ethiopia as part of the Longitudinal Infant Growth and Development (LIDG) study (NCT06296238). We describe adaptations and strategies to address and optimize data capture (e.g., dry electrode tips to improve scalp contact, tailored protocols, and adequate equipment specifications), environmental challenges (e.g., space constraints, lack of water supply, power outage) and cultural factors (e.g., hair type) unique to the study setting and population. Our formative research underscored the importance of creating awareness among community members (e.g., mothers, fathers, and religious leaders) and local clinicians to improve community engagement and buy-in. Culturally sensitive child behavior management techniques were also critical to ensure EEG completion and high data quality. With community sensitization, we had high consent rates for EEG/VEP (>90%). We completed EEG recordings within an average ± standard deviation of 20 ± 11 minutes. After data processing, approximately 90% and 70% of participants met predefined data quality thresholds for resting EEG and VEP, respectively. Implementing mobile EEG/VEP was feasible and acceptable in rural Ethiopia, with a relatively high proportion of recordings meeting quality standards.

Keywords: electroencephalography (EEG); low- and middle-income countries (LMIC); neurodevelopment; nutrition; visual evoked potential (VEP).

Figure 1

Enobio-32 Drytrodes (left) and Foretrodes (right) (Neuroelectrics, Barcelona, Spain). Source: https://www.neuroelectrics.com/solutions/enobio/32.

Figure 2

Use of tubular net bandage to provide tighter head cap fit.

Figure 3

Modified EEG set-up at a local health center from the child’s viewpoint.

Figure 4

Comparison of power spectral density of individual EEG signals acquired on different power sources. (A,B) Two individual examples when EEG recording was powered by a generator. (C,D) Two individual examples when EEG recording was powered by grid. (A–C) were collected on the same day; (D) was collected one day after the other recordings. All data was collected at one of the study health centers. Each line in each subfigure represents a signal in an electrode. Topomaps are plotted for the frequency of line noise in each subfigure.

Figure 5

Power spectral density from resting-state, eyes-opened EEG averaged across electrodes in two regions of interest (N = 173).

Figure 6

Grand average VEPs from three occipital electrodes (O1, O2, and Oz). Data were collected from 165 LIDG participants by non-specialist study nurses (top) and two adult volunteers (three recordings per volunteer) during pilot testing by EEG experts in Boston (bottom).