A service based on AI algorithms that identifies potential foci of epilepsy with 99% accuracy in 5 minutes
How does it work?
1. The doctor uploads EEG data in EDF / EDF + C format.
2. The application analyzes them and marks the generalized discharges.
3. The obtained data is analyzed and confirmed by a specialist.
The service helps doctors to
Make more effective medical decisions through preliminary sorting and visualization of areas with pathologies
Accelerate research analysis and diagnosis
Prioritize research with potential to have pathologies
Reduce the risk of medical errors
Detailed operating principle
Interictal epileptiform discharges (IEDs) on the EEG are one of the key indicators of potential epilepsy. This type of brain activity, seen outside of seizures, is important for diagnosis. With the help of localization of focal discharges in the EEG channels, it is possible to predict the localization of the epilepsy focus.
The model is based on a predictive decision tree ensembles. This is a machine learning algorithm that classifies images, identifying deviations and their most characteristic areas. It marks the discharge accurately enough to make a preliminary conclusion that can be further analyzed by a specialist.
High accuracy of the model: sensitivity (proportion of correctly identified positive results) - 51%, specificity (proportion of correctly identified negative results) - 99%
Fastest solution on the market: marking a 10-hour recording takes up to 5 minutes instead of 3 hours when manually processed by a specialist
Ease of use