Cross-cultural Evaluation of Dementia Passive BCI Neuro-biomarker Candidates

A successful unsupervised clustering methodology implementation, taking into account cross-cultural effects in neurophysiological datasets, in a so-called ‘AI for social good’ environment shall allow for an efficient dementia digital neuro–biomarker development for early-onset prognosis of a possible cognitive decline in many countries around the world. We present an encouraging initial study of a cross-cultural EEG data collection in Japan and Poland using the same experimental paradigm of learning facial emotion and evaluating elderly participants with various cognitive decline stages. We also compared two wearable EEG devices and derived theta-band fluctuations. Next, we cluster the obtained fluctuation features using an unsupervised technique of uniform manifold approximation and projection (UMAP) in the facial emotion video-clip judgment learning and evaluation sessions with Japanese and Polish elderly participants. In the presented pilot study, we report findings from thirty-five Japanese and twenty-six Polish elderly volunteers instructed to learn to evaluate facial emotions. The documented pilot project showcases vital social and cross-culturally evaluated methodology of artificial intelligence (AI) application for an early-onset dementia prognosis.