Functional near-infrared spectroscopy for human brain age group classification by machine learning

Abstract

Aging brain undergoes multiple structural and functional changes. These may contribute to an increased risk of neurodegenerative disease (NDD) and other age-related diseases, highlighting the importance of assessing deviations from healthy brain aging trajectory. In this human brain study, 50 healthy adults were investigated by functional near-infrared spectroscopy (fNIRS). A resting state single channel multiwavelength fNIRS was measured from the forehead in a supine position. The subjects were divided into four age groups. A machine learning approach was utilized for age group classification by using support vector machine and random forest learners with nested cross-validation. The results suggest brain aging effects being more distinct in the oldest age group and a difference in the brain aging for the subjects of the in-between groups. Our study shows high potential for the use of fNIRS in the analysis of brain aging.