Research
A short introduction to the program. For a specific paper, see all publications, by topic.
Reading EEG takes scarce expert time. MORGOTH is a single model for comprehensive automated interpretation, developed and validated at multiple centres.
We measured interrater reliability among expert electroencephalographers identifying seizures and rhythmic and periodic patterns, and the noise in expert diagnosis of epilepsy.
SpikeNet is a deep learning model for automated detection of interictal epileptiform discharges in EEG.
A brain age estimated from one night's EEG, and a sleep-derived brain health score studied against long-term cognitive decline.
Dynamic instability of breathing during sleep, quantified from routine recordings, predicts cognition, disease and mortality.
Automated scoring of clinical sleep studies, so that large sleep datasets can be analysed consistently.
EEG patterns and their prognostic significance in a multicentre cohort of 1,000 patients, and an international EEG database for outcome prediction research.
Clinical EEG, sleep and ECG datasets released for reuse, including the Human Sleep Project and the Harvard-Emory ECG Database.
Methods that extract diagnoses and outcomes from clinical notes, such as cognitive impairment and functional outcome after cardiac arrest.