액티브 시니어를 위한 머신 러닝 기반 거동 분석 시스템 구축

초록

Gait impairment is one of the main symptoms of neurodegenerative diseases such as dementia, Parkinson’s disease, etc. For this reason, lots of previous studies tried to develop new methodologies based on statical analysis for predicting brain diseases. Statistical analysis is a good choice for solving most engineering problems. However, neurodegeneration patients cannot wait for progression because of their limited time. In this study, we focused on analysis time reduction. We acquired ten sets of the gait sample by Arduino pro micro using the MPU6050 accelerometer. The sampling frequency was 200Hz, and data were acquired on the x, y, and z acceleration, pitch, roll, and yaw. The raw data sets were pre-processed to 100 normal and 60 abnormal gait data, where ten were used in the test, and the others in the study of the machine. The machine learning achieved an 80% total accuracy at the end of this study.

키워드