Posture and gait analysis for research on risk of falling for older adults

As long as the population is aging, more and more eyes are drawn to the aging related problems. Falls and its induced problems are annoying the older people and are considered as a heavy burden no matter to the individuals or to the whole society. Hence, to prevent falls for the aged society is extremely important and necessary. The objective of this research aims at developing a falls predicting and analysis framework based on biomechanical principles and machine learning classification techniques.;To fulfill this goal, people from different ranges of age have been invited to provide their performance in the Reduced Gravity and Biomechanics (RGB) Lab in New Mexico State University. A motion capture system is considered to be an ideal equipment for this laboratory setting, combined with an instrumented treadmill. The integrated system is capable of obtaining the whole-body motion information of individuals, including kinematics and kinetics data.;Standing on the achievements of myriad scholars and researchers, the existing risk factors of falls and falls predictors, which are considered to be fall features, will be reviewed and analyzed. Those widely prevalent fall features have been validated through the motion data from the participants invited to this research. The participants were labeled according to their distinguishable performance of these features, and the most distinctive features have been verified.;Compared to the traditional falls predicting assessments, it is the tendency that biomechanical models provided more objective and scientific support. Therefore, the inverted pendulum models of human walking introduced in this research, and are prospected to be the tool to derive novel falls predictors. Also, the newly derived falls predictors have been analyzed for their effectiveness.;So long as the boom of the modern computers' performance, machine learning (ML) techniques are introduced to more and more fields. And this research happens to be a perfect application area for ML techniques. By employing five different ML algorithms, five falls predictive models have been developed and studied. By comparing and analyzing them with each others, the feasibility of the application was verified. The out compete predictive model was given according to their predicting accuracies.;In summary, this research concentrates on the application of new techniques in an underdeveloped area, trying to set up a framework for the purpose of falls prediction for the older population. The highlight of this research is the introduction of biomechanics and machine learning techniques which give the society a new tool to solve the problem. However, as the research is still in its early stage, various problems, improvements and discussions were given.