DigRTEpi was designed to evaluate EEG compatibility with the fitness to drive in adults and to screen for IED-induced cognitive deficits that are relevant to daily social functioning in people with epilepsy from infancy (8 years and older) to adulthood. The artificial intelligence used in DigRTEpi consists of an advanced visualization technique that calculates the new information content or entropy of a window that progressively slides across the EEG while it is recorded. A deep neuronal network classifies each window’s content in real-time and triggers either an obstacle in the driving videogame or a video in the neuropsychological bedside test. Our machine learning algorithm is currently the fastest method for classifying brain bio signals and triggering a stimulus in science and the market, with a mean prediction time for IEDs of 98 milliseconds. TThe very good prediction performance on EEGs from three different university hospitals in Europe and the U.S. and for different epilepsy types and syndromes, indicates good generalizability of the artificial intelligence. In addition, the IED detection algorithm can be personalized. The driving videogame measures virtual accidents and reaction times. We correct for the digital latency required for a visual stimulus to appear on the screen of a computer and measure effective reaction times from the time the stimulus can be perceived by a subject, rather than from the time the stimulus is triggered. This is important because the time between IED detection, which triggers a signal, and on-screen stimulus appearance differs from one trigger to the next. Calculating mean or median digital latencies and subtracting them from individual reaction times, as most digital correction methods do, would overestimate, or underestimate the variability of digital latencies and thus the actual reaction times of the subjects. The neuropsychological bedside test called interictal Automated Response Test (iART) was developed to detect IED effects on daily social functioning. iART measures transitory cognitive impairment during brief IEDs in 17 different native language in an objective and standardized manner. Multilingualism will improve sensitivity and specificity of detecting subtle cognitive deficits during short IEDs if subjects are tested in their native language. In addition, it increases the probability that iART will be disseminated internationally. 38 videos with short questions cover all essential neuropsychological domains for daily social functioning in each language. DigRTEpi preserves video-EEG synchronization when EEG data is exported and opened with other programs. This can be advantageous when EEG recordings are sent for second opinion between hospitals that often do not have the same EEG systems. DigRTEpi was tested in a pilot study with patients. IED prediction performance in each new patient’s EEG was comparable to the performance of detecting IEDs in the test segments of the EEG training data set. From a medical perspective, generalized IEDs appear to affect frontal and temporal lobe functions to varying degrees, or in other words, different brain regions have different susceptibilities to generalized IEDs. DigRTEpi has been disseminated at the American Epilepsy Society meetings in 2019 and 2021, at a workshop at the European Epilepsy Congress in 2022, in scientific publications, and on a website (see URL below). Public engagement was compromised due to COVID-19. Exploitation will occur through an intellectual property agreement and integration of DigRTEpi into clinical practice at Goethe University Frankfurt and Yale University. Funding from Distr@l, a German government program that together with the EU promotes knowledge and technology transfer to European interest groups will support DigRTEpi’s further development.
