Doose and Lennox-Gastaut (syndromes) are rare generalized electroclinical affections of early infancy of variable prognosis which manifest with very diverse kinds of seizures. Very frequently, these types of epilepsy become drug resistant and finding reliable treatment results is very difficult. As a result of this, fighting against these syndromes becomes a long term (or endless) event for the little patient, the neurologist and the parents. A lot of Electroencephalographic (EEG) records are so accumulated during the child's life in order to monitor evolution and correlate it with medications. So, given a bunch of EEG, three questions arise: (a) On which year was the child healthier (less affected by seizures)? (b) Which area of the brain has been the most affected? (c) What is the status of the child with respect to others (which also have a bunch of EEG, each)? Answering these interrogations by traditional scrutinizing of the whole database becomes subjective, if not impossible. We propose to answer these questions objectively by means of time series entropies. We start with our modified version of the Multiscale Entropy (MSE) in order to generalize it as a Bivariate MSE (BMSE) and from them, we compute two indices. All were tested in a series of patients and coincide with medical conclusions. As far as we are concerned, our contribution is new.
Keywords: EEG dynamics; Time series entropy; drug resistant children epilepsy; epileptic encephalopathies.