Viability of an EEG‐based Automatic System for Early Dementia Screening

Background Based on recent demographic projections the worldwide number of people with dementia will triple by the year 2050, a socio‐economic burden that will disproportionally affect low‐ and middle‐income countries (Nandi et al., 2022). Therefore, a reliable, objective and cost‐effective screening methods are needed to be able to apply early interventions and slow down a decrease in quality of life for millions of patients. To address this issue, we explore viability of an EEG based system as an affordable solution for early dementia screening. Method The tested system includes a mobile amplifier (mBrainTrain) with 24 channel sponge EEG cap (Greentek) and a custom software (BDI) for EEG data acquisition and analysis (BrainTrip). The BDI software automatically detects and removes common EEG artifacts, calculates spectral EEG features known to be associated with cognitive decline (Sanchez‐Reyes et al., 2021), and uses density clustering and a multi‐layer perceptron classifier to predict the cognitive ability using a score between 0 and 100. We tested 448 senior participants who undertook five cognitive screening tests (MoCA, ADAS‐cog, ACE‐III, Eurotest, Phototest) and eight minutes resting state EEG recording on three separate sessions. Result Our previous analysis was performed on about 50% of the total included subjects, which we now expended to the full dataset. Additionally, more provisions were made to prevent potential overfitting in building the final predictive model. Setting the cut‐off value of the BDI at 50 points, the specificity was at 97% and sensitivity at 65% compared to a consensus of all other cognitive tests. Assuming a dementia prevalence of 10%, the BDI test achieved an accuracy of 94%. Conclusion Based on the high accuracy of the EEG‐based assessment in detecting cognitive decline and dementia, it could be a promising option for dementia screening. It is an efficient, low‐cost and non‐invasive option for early intervention, especially in a primary healthcare setting.