Artificial intelligence based algorithm development for improved diagnostics using electroencephalography
Electroencephalography (EEG) reliably predicts outcome in coma after cardiac arrest (CA) (use case [UC] 1) and is promising for prehospital detection of a brain bleed (UC 2). State-of-the-art scalp EEG requires specialized personnel for application and interpretation. Moreover, interpretation is time-consuming and prone to interrater differences limiting application of EEG in both UCs. The consortium aims to develop artificial intelligence (AI) algorithms for personalized decision support in both UCs based on subhairline EEG, and integrate these algorithms into a simple-to-use EEG recording device developed by TrianecT (together: SMART-EEG).
In the Netherlands, ±5000 patients are hospitalized in coma after CA, and ±7500 people suffer a brain bleed each year. For UC1, SMART-EEG enables early application of EEG by intensive care unit nurses instead of EEG technicians. Combined with AI for interpretation, the predictive value of EEG for outcome prediction is expected to increase significantly. This will shorten hospital stays and improve long-term outcome. For UC2, SMART-EEG is expected to reliably detect a brain bleed in the prehospital setting. This will reduce time-to-treatment leading to improved patient outcome (time=brain). The potential savings by successfully implementing SMART-EEG are approximately ±€52 million per year.
The consortium will first prepare AI-algorithms based on existing scalp EEG data. New subhairline EEG data will be collected in 225 coma patients and 600 patients with a suspected stroke, among which 100 with a brain bleed. These subhairline EEG data will be used to adapt the AI-algorithms specifically for personalized decision support based on subhairline EEG.
We will primarily deliver two AI-algorithms: 1) for prediction of neurological outcome in coma after CA and 2) for the detection of a brain bleed among patients with a suspected stroke. These algorithms will be integrated into the TrianecT EEG recording device. SMART-EEG will be demonstrated in a clinical setting for both UCs.
