Epilepsy stands as one of the major neurological disorders which causes sudden seizures through abnormal brain electrical activity thus producing dangerous health risks as well as psychological effects and social challenges. The use of EEG and video-EEG diagnostic methods outside clinical environments faces challenges in detecting seizures both at early stages and continuously especially during sleep or outside clinical settings. The need for smart wearable solutions to monitor and analyze patient movements in real time with accuracy has increased because of these challenges. Our research presents a smart wrist-worn system which integrates an ESP32 microcontroller with an MPU9250 inertial measurement unit (IMU) that includes an accelerometer and a gyroscope and a magnetometer. The system converts motion data into images by applying wavelet transform processing before a convolutional neural network (CNN) identifies motor seizure-related spastic movements from standard activities including walking, running and climbing stairs. The idea is to improve detection accuracy and reduce false alarms, enhancing patient safety and self-management skills. This study addresses these issues by developing and evaluating a new wrist-worn seizure detection device. Wrist-worn seizure detection, which makes use of accelerometers, gyroscopes, and magnetometers. These are connected to an ESP32 processor which sends data to a dashboard. The device was tested on five volunteers to elicit seizure related tremors as well as data on natural movement patterns such as walking, running and climbing stairs. The system developing a kinematic analysis for the data which enhances the accuracy of the seizure classification for faster intervention and improvement of patient safety.