Implementation of K-Nearest Neighbor for Fall Position Detection of Dementia Patients Based Microcontroller

Abstract

A microcontroller-based detection tool for the presence of patients with dementia has been made using the K-Nearest Neighbor (KNN) method with the help of coordinate points that can be seen via Google Maps. which is based on patient care with a patient-oriented approach. The targets of this research are (a) designing and implementing a fall detection system using the mpu6050 sensor, (b) using the (KNN) method to determine the coordinates of the location of dementia patients using GPS. The research method starts from making a prototype and measuring system performance. The test results on GPS produced an average latitude error of 0.002091% and an average longitude error of 0.000032% in Pauh District, while in Lubuk Kilangan District the average latitude error was 0.002641% and an average longitude error of 0.000150%. The KNN method with the Eucledian distance formula can help supervisors find out the nearest police station to the patient through the coordinate points detected by GPS by taking the smallest value from the comparison of values in the form of degrees between the Pauh police station and the Lubuk Kilangan police station for the patient. Overall the tool can function well.