인공지능을 적용한 정적자세의 분석

초록

We projected that Static Balance Ability(SBA) was measured as the ground truth of Artificial Intelligence(AI) model from Quiet Stance(QS) feature of status were extracted, modeled with AI, and verified by statistical methods. For this study, healthy adult men(N=20, 22.9±1.88) participated. While maintaining QS, the displacement of Center Of Pressure(COP) occurring under both feet were measured on the x (mediolateral) and y(anteroposterior) axes using sensor insole. Superior(n=5) and Inferior groups(n=5) were divided as SBA(one leg stance with closed eyes) which was set to Ground Truth, and the status of QS was modeled with BinaryClassifier AI of based on the feature of each participant's QS. Mean, standard deviation, and frequency distribution of COP displacement of each Superior and Inferior group were calculated. In addition, the ratio of the stability and instability status which were output from AI model of the evaluation group(n=10) was tested with the Independent variable t-test. And, correlation analysis(Pearson, both side) was performed with SBA and the ratio of stability and instability on evaluation group(p<.05). As results of study, Superior SBA group maintained QS by increasing mediolateral displacement of dominant right foot. AI modeling from S-4 and S-5 of Superior group with SBA, while maintaining a QS in evaluation group, ratio of stability of Superior SBA group higher than Inferior SBA group. And the ratio of instability of Inferior SBA group was higher than Superior group with SBA(p<.05). There was no significant correlation between SBA of evaluation group and the ratio of stability and instability output from AI modeling(p>.05). In conclusion, when the BinaryClassifier AI model was set SBA as Ground Truth, it was possible to determine the two status of good or bad QS.

키워드