Data analysis for human balancing

Abstract

Human Balance ability is one of human’s basic physiological functions to ensure standing, walking and completing complex actions. The investigation about human balancing is very important in the areas of medical diagnosis, sports, and so on. Data analysis for human balancing is the core of balancing tests and human balance ability assessment. we have summarized different kinds of sensors for fall detection and methods of vision-based detection. The conclusion is that force distribution under human feet is a very helpful indicator of human balance ability when they trying to balance themselves. Therefore, an optics-based sensor, called FTIR (Frustrated Total Internal Reflection) sensor, has been refined, calibrated and used to measure the pressure under human feet in evaluating the balance ability of the drunk and the elderly persons. The relationship between the pressure and pixel intensity of the FTIR sensor has been calibrated, the pressure distribution variation can be quantitatively determined. To explore the quantitative evaluation of pressure on balance ability of human body, data collection is the basis of data analysis. To collect data of human balancing, several experimental protocols under different conditions have been provided. Features extraction is the second step of data analysis. From two simple and effective parameter indexes, that are CoP (Center of pressure) and CoG (Center of Gravity), quantitative evaluation features , that are temporal and frequential feature,of balance ability have been extracted. The CoP is a kinematic parameter index, the CoG is a dynamic parameter index. From pressure distribution variation, the calculation method of CoP, CoG and other quantitative evaluation features have been studied. Three classification methods and k-means clustering method of machine learning, result evaluation methods of statistics have been analyzed. The evaluation results under different conditions have been compared with several experimental conclusions. Hence, a whole architecture of data analysis of human balancing has been presented. We have proved that the CoP of the group before for the degree of Master of Philosophy at The University of Hong Kong in December 2023 and after drinking was changed and that the balance ability of the drunk person was Key words: FTIR sensor; Pressure; Balance Ability; Quantitative Evaluation; generally lower than the CoP value of the healthy group under normal circumstances. Then we have done research on the fall risk assessment applicable to the elderly of the pressure of 168 subjects to evaluate their balance ability based on the FTIR balance sensor data. In addition to comparing the results at the behavioral and physiological levels with those from the balance sensor, we also have compared the relationship between the balance sensor outcomes and those at the functional level.