Design of novel AI algorithms, apps and devices for epilepsy management

Abstract

Epilepsy is a chronic neurological disorder characterized by recurrent, unprovoked seizures, affecting over 50 million people worldwide. While anti-seizure medications (ASMs) remain the primary treatment, approximately 20–30% of individuals develop drug-resistant epilepsy, leaving them vulnerable to uncontrolled seizures, cognitive impairments, and an increased risk of Sudden Unexpected Death in Epilepsy (SUDEP). The unpredictable nature of seizures significantly impacts the quality of life, often leading to social isolation, anxiety, and difficulties in daily activities. This highlights the urgent need for real-time seizure detection and self-management tools to enhance patient safety and improve long-term outcomes. This thesis presents a multi-faceted approach to epilepsy management by integrating wearable technology, artificial intelligence (AI), and digital health solutions. A wearable seizure detection system in the form of a smartwatch has been developed to continuously monitor physiological parameters such as movement patterns, heart rate variability, electrodermal activity, and temperature fluctuations. Using AI-driven algorithms, the smartwatch detects seizure events in real time and alerts caregivers, enabling timely intervention. To complement this system, a mobile-based epilepsy self-management application, Epilepto, has been designed. This application allows users to log seizure episodes, track potential seizure triggers, receive personalized medication reminders, and share health data with clinicians and caregivers, fostering a more proactive approach to epilepsy care. Recognizing that some individuals may prefer non-wearable solutions, this research also introduces a non-contact, AI-driven video-based seizure detection system. Leveraging deep learning techniques, this system analyzes body movements from video recordings to identify seizure events with high accuracy, offering an alternative for individuals who may not tolerate wearable devices. Additionally, to address the challenge of medication adherence, a smart pillbox has been developed, which provides automated reminders and real-time tracking of medication intake, reducing the likelihood of missed doses and breakthrough seizures. By integrating AI-driven wearable and non-wearable seizure detection systems with digital health interventions, this work presents a comprehensive, patient-centric framework for epilepsy management. These innovations aim to empower individuals with epilepsy, enhance real-time monitoring and early intervention, and ultimately improve clinical outcomes, safety, and quality of life. The findings of this research contribute to the growing field of digital health and highlight the potential of AI-powered assistive technologies in the management of chronic neurological disorder.