A direction-sensitive fall detection system using single 3D accelerometer and learning classifier

The rate of fall incidence among the elderly people is ever increasing. It is at the sixth position in the list of causes of death for the people aged between 60 and 65; the second between 65 and 75; the first over 75. Treatment of a patient, experiencing complications due to a fall, within the first 12 minutes after a fall brings a survival rate of 48% –75%. So, fast and accurate detection of fall events is emerging as a big necessity for many countries, especially for the advanced world where the society adopts the culture of independent living for elderly people. It is also important to determine the direction of a fall as it can help determine the locations of joint weakness and fracture quickly. Researchers' claims of fall detection accuracy of over 90% are based on accelerometers and embedded extra sensors like gyroscopes, cardio tachometer, magnetometer, and barometric pressure sensors. However, most such fall detection algorithms have been developed based on observational analysis of the data gathered, leading to thresholds settings for fall/non-fall situations. To detect the direction of fall, some researchers uses gyroscope or more accelerometers. The proposed method, using single 3D accelerometer and machine learning algorithm particularly SVM (Support vector machine) is to detect 4 types of falls (forward, backward, right and left). When applied to experimental data from 13 male subjects, the proposed system discriminates between falls and activities of daily living (ADL) with better than previously reported accuracy level. The system is reliable, user friendly and cost effective.