Investigation of gait features for stability and risk identification in elders.

Today, eldercare demands a greater degree of versatility in healthcare. Automatic monitoring devices and sensors are under development to help senior citizens achieve greater autonomy, and, as situations arise, alert healthcare providers. In this paper, we study gait patterns based on extracted silhouettes from image sequences. Three features are investigated through two different image capture perspectives: shoulder level, spinal incline, and silhouette centroid. Through the evaluation of fourteen image sequences representing a range of healthy to frail gait styles, features are extracted and compared to validation results using a Vicon motion capture system. The results obtained show promise for future studies that can increase both the accuracy of feature extraction and pragmatism of machine monitoring for at-risk elders.