Sleep staging is a fundamental step in diagnosis and treatment of sleep disorders. However, it is a time-consuming and tedious task, which has to be performed by medical experts.
Hypnograms (graphs that represent the stages of sleep as a function of time) are usually obtained by visually scoring the recordings from electroencephalogram, electrooculography and electromyography. The data is heterogeneous as there is a multiplicity of physiological parameters recorded at night (using a wired connection which may be a source of disturbance for the patient’s sleep) and rules of interpretation (on the basis of the American Academy of Sleep Medicine – AASM - textbook).
There is a need for integrating and analyzing information from heterogeneous data sources with high accuracy in order to be able to automatically read hypnograms.
The solution is a novel system that mimics visual decision making process of clinical sleep staging using symbolic fusion and an evolutionary algorithm (for adaptive thresholds), with an approach based on AASM guidelines.