Frontiers in Neuroscience, Год журнала: 2025, Номер 19
Опубликована: Апрель 30, 2025
Manual scoring of longitudinal electroencephalographical (EEG) data is a slow and time-consuming process. Current advances in the application machine-learning artificial intelligence to EEG are moving fast; however, there still need for expert raters validate data. We hypothesized that power-frequency profiles determining state 'set framework' communication between neurons. Based on these assumptions, method with set frequency profile each vigilance state, both sleep awake, was developed validated. defined seven states functional brain unique terms frequency-power spectra, coherence, phase-amplitude coupling, α exponent, excitation-inhibition balance (fE/I), aperiodic exponent. The new requires manual check wake-sleep transitions therefore considered semi-automatic. This semi-automatic approach showed similar exponent fE/I when compared traces scored manually. faster than scoring, advanced outcomes were stable across datasets epoch length. When applying neurexin-1α (Nrxn1α) gene deficient mouse, model synaptic dysfunction relevant autism spectrum disorders, several genotype differences 24-h distribution detected. Most prominent decrease slow-wave comparing wild-type mice Nrxn1α-deficient mice. methodology puts forward an optimized validated analysis pipeline identification translational electrophysiological biomarkers disorders related architecture E/I balance.
Язык: Английский