Evaluation of fall detection for the elderly on a variety of subject groups

Falls in the elderly are a major problem for today's society. If the elderly could get help immediately after the fall, the severity of the injury could be reduced. Also, it results in decreasing the rate of death and the medical cost. This paper presents a fall detection algorithm based on the threshold value of the maximum peak resultant acceleration to classify falls and Activity of Daily Live (ADL). Two types of the experiments were investigated. Type A) ten young subjects performed both falls and ADL. Type B) ten young subjects performed falls whereas ten elderly subjects performed ADL. In the experiment, tri-axial accelerometer was mounted on the trunk. There were four categories of falls: forward fall, backward fall, left and right side fall and six categories of ADL: sit-stand, stand-sit, sit-lie, lie-sit, bend down, and walking 2 m. For the threshold of the maximum peak resultant acceleration at 1.9g, falls could be distinguished from ADL with 100% sensitivity in both Type A and B while specificity for Type A and B were 96.11% and 98.33%, respectively. Results indicate that the trend in classification of fall from ADL in the elderly could gain the increase in error. Therefore, more sophisticated algorithms for the classification of fall from ADL in the elderly are needed to improve performance of detection.