A sustainable artificial-intelligence-augmented digital care pathway for epilepsy: Automating seizure tracking based...

Area of Science:

• Neurology
• Artificial Intelligence
• Digital Health

Background:

• Scalp electroencephalograms (EEGs) are vital for neurological evaluations, especially in epilepsy diagnosis.
• Limited specialized expertise for EEG interpretation hinders widespread access to care.
• Existing artificial intelligence (AI) models partially address EEG analysis, but a fully automated system is needed for routine interpretation.

Purpose of the Study:

• To develop an AI-augmented model for automated EEG seizure tracking.
• To support a sustainable digital care pathway for epilepsy (DCPE).
• To improve patient monitoring, decision-making, and compliance within the DCPE.

Main Methods:

• Proposed an AI-augmented framework utilizing machine learning for quantitative EEG analysis.
• Employed random forest with principal component analysis, support vector machines with KBest feature selection, and convolutional neural networks.
• Conducted focus group discussions with healthcare professionals to assess feasibility, usability, and sustainability.

Main Results:

• Achieved high accuracy rates: 96.52% (random forest), 95.28% (SVM), and 97.65% (CNNs).
• Automating the diagnostic process can significantly reduce epilepsy diagnosis time.
• Sustainability is contingent on infrastructure, personnel, training, digital literacy, funding, and regulations.

Conclusions:

• The AI-augmented system enhances epilepsy management through accurate seizure tracking and improved monitoring.
• Facilitates collaborative decision-making and promotes patient-centered care.
• Contributes to a more sustainable digital care pathway for epilepsy.