Artificial neural network trained on smartphone behavior can trace epileptiform activity in epilepsy

Area of Science:

• Neuroscience
• Biomedical Engineering
• Digital Health

Background:

• Epilepsy is characterized by epileptiform discharges, abnormal electrical brain activity.
• Implanted devices can monitor these discharges and deliver neurostimulation.
• Continuous recordings offer insights into behavioral correlates of epileptiform discharges during daily life.

Purpose of the Study:

• To investigate the relationship between daily digital behavior and epileptiform discharges in epilepsy patients.
• To develop a personalized model for tracking epilepsy disease activity using smartphone data.

Main Methods:

• Collected electrographic recordings and smartphone touchscreen interaction data from eight epilepsy patients over 35,714 hours.
• Utilized an artificial neural network model to analyze if smartphone behavior reflected epileptiform discharges.

Main Results:

• Smartphone behavioral inputs generated personalized model outputs that correlated well with observed electrographic data (median R=0.4).
• Demonstrated realistic reconstructions of epileptiform activity based on smartphone usage patterns.

Conclusions:

• Day-to-day digital behavior, captured via smartphones, can be transformed into personalized markers of epilepsy disease activity.
• This approach offers a novel, non-invasive method for monitoring epilepsy progression and management.