Validity of Awareness of Epileptic Seizures in Patients with Severe Motor and Intellectual Disabilities

Abstract

　Epileptic seizures in patients with severe motor and intellectual disabilities (SMID) are difficult to distinguish and may cause respiratory depression; therefore, early detection is crucial. We explored the feasibility of using artificial intelligence technology to detect epileptic seizures early, leveraging facial images and pulse rate data from patients with SMID. The research participant, a patient with SMID diagnosed with Oshima’s classification 1, often exhibited increased pulse rate and decreased transcutaneous oxygen saturation during epileptic seizures. We categorized facial expressions and pulse rate data into three scenarios: days without seizures (A), preseizure (B), and postseizure (C). Using results from previous multiple regression analyses, we performed logistic regression and neural network analyses. The data across the three scenarios were accurately classified as “no seizures,” “possibility of seizures,” and “seizures present” even when using these analyses.