Sparse Multiple Instance Learning for Elderly People Balance Ability

Yuqi Zhang, Duo Su, Xiaoxi Zhao, Yingjie Tian

Abstract: Falling occurs when a person’s body (excluding their feet) accidentally touches the ground, resulting in severe injury for the elderly. As a response, determining old people’s balance characteristics and assisting them in preventing falls is important. We proposed the sparse SyMIL with SCAD to find the important factors that may cause an elderly person to fall, which can yield realistic and accurate results. This work can also be used in future research about falling prevention and balance supervision.