A Continuous Clustering Algorithm for Detection of Local Sleep in Humans

What is it about?

Work in animal models has demonstrated the occurrence of "local sleep while awake" - islands of neurons showing inactivity identical to sleep while the animal is clearly awake, with the volume of such inactivity increasing with time awake. The authors previously showed evidence for local sleep occurring in humans from analysis of intracranial recordings of brain electrical activity using a k-means clustering binary classification scheme. The current paper is a re-analysis of the same data using a continuous fuzzy c-means metric for classification. The analysis revealed focal changes again consistent with the local sleep hypothesis, and also demonstrated roughly equal contributions to the classification state from the frequency and amplitude of the signal.

Featured Image