Artificial intelligence-driven closed-loop devices in sudden unexpected death in epilepsy prediction and prevention: Insights from persons with epilepsy and caregivers.

OBJECTIVE

The absence of strategies for predicting and preventing sudden unexpected death in epilepsy (SUDEP) is intertwined with the lack of studies measuring users' attitudes toward potential innovative interventions. The NEUROSENSE Project (http://www.neurosense-project.eu) aims to evaluate novel SUDEP-predictive neuroendocrine biomarkers in interstitial fluid. The ultimate aim is to develop an artificial intelligence-driven closed loop device (AI-CLD) prototype that can recognize life-threatening seizures and prevent SUDEP through automatic intervention. The current study introduces the potential use of AI-CLDs in SUDEP prediction and prevention, while assessing person with epilepsy (PWE) and caregiver (CG) attitudes toward AI-CLD adoption and implementation.

METHODS

A qualitative study was conducted through three focus groups involving PWEs and CGs. Participants were recruited through the NEUROSENSE Patient Advisory Board, with discussions facilitated through a semistructured interview guide. The study followed grounded theory and qualitative content analysis methods. Data were collected between October 2024 and February 2025, with all sessions transcribed and analyzed.

RESULTS

Three main areas emerged from the analysis: expectations of AI-CLDs for SUDEP prediction and prevention, decision-making processes involving AI use in health care, and barriers and facilitators to AI-CLD adoption. PWEs and CGs generally expressed positive attitudes toward AI-CLDs, supporting automatic data sharing with health care providers and real-time alerts. However, concerns about AI accuracy, overreliance on automation, and the need for control over interventions were raised. Both groups preferred wearable devices over implanted solutions, emphasizing comfort and discretion as critical factors for adoption.

SIGNIFICANCE

This study highlights the potential of AI-CLDs in improving the prediction and prevention of SUDEP, showing promise for enhancing patient safety through real-time monitoring and interventions. The findings underscore the importance of user-centered design in device development, emphasizing comfort, control over interventions, and integration into daily life. This research provides insights useful for future development aiming to improve PWE and CG confidence in using AI technologies for epilepsy care and risk management.