ABSTRACT
Seizure manifestations may be difficult to describe in words alone. Thus, initially, 24 seizure images were developed to support communication and gain assistance during obtaining the patient's history. Before being used in clinical practice, these seizure images must be investigated for validity and reliability. We tested the images with untrained participants including patients with epilepsy, persons who had witnessed seizures, and participants who had neither had nor witnessed epileptic seizures. The participants filled in a questionnaire evaluating the images twice within 3â¯days. The participants were asked to choose one of the 2 written descriptions that best matched each seizure image. The validity was assessed using one-proportion z-test. The reliability was assessed by Gwet's AC1. The first analysis showed that the proportion of correctly identified seizure images was higher than 70%, except for 2 images representing dystonia and myoclonus. The dystonia image was modified, and the myoclonus image was removed. In the final evaluation, the seizure images were identified with an overall correctness ratio of 96%. The final AC1 of the seizure images was classified as very high. The final 23 seizure images are proved to be valid and have a high agreement that can be used in clinical practice.