Start Date
18-6-2025 12:30 PM
End Date
18-6-2025 2:00 PM
Abstract
This study aims to validate the validity of the AI-based physical function testing system (myFitNote®) using pose estimation technology through the Timed Up and Go (TUG) test. To achieve this objective, validity verification was conducted using values measured by experts, which are known as the traditional method. The study participants consisted of 52 individuals aged 60 and above. Data processing was performed using IBM Statistics 20.0 and the R package 'blandr' to calculate the mean and standard deviation. For validity verification, correlation analysis (r) and Bland & Altman Plot were conducted. A strong correlation (r = 0.958) was observed between the values measured by the experts and those obtained using the app-based digital measurement system(myFitNote®). The results from the Bland & Altman analysis showed stable confidence intervals, confirming high agreement between the two methods. The absolute error rate was found to be 5.78%. Consequently, the validity of the TUG test between the expert method using a stopwatch and the AI-based physical function testing system (myFitNote®) was confirmed to be high.
Recommended Citation
choi, muncheong CEO,Ph.D.; song, Hong-sun Prof.; Kim, Alchan Prof.; Kim, Da-Ae Dr.; Kim, Ji-woong PhD candidate; and Jeoung, Bogja Prof., "Validation of AI-based Physical Function Test (myFitNote) in Timed Up and Go (TUG) Test" (2025). International Symposium of Adapted Physical Activity and International Symposium on Physical Activity and Visual Impairment and Deafblindness. 39.
https://sword.mtu.ie/isapa/2025/day3/39
Validation of AI-based Physical Function Test (myFitNote) in Timed Up and Go (TUG) Test
This study aims to validate the validity of the AI-based physical function testing system (myFitNote®) using pose estimation technology through the Timed Up and Go (TUG) test. To achieve this objective, validity verification was conducted using values measured by experts, which are known as the traditional method. The study participants consisted of 52 individuals aged 60 and above. Data processing was performed using IBM Statistics 20.0 and the R package 'blandr' to calculate the mean and standard deviation. For validity verification, correlation analysis (r) and Bland & Altman Plot were conducted. A strong correlation (r = 0.958) was observed between the values measured by the experts and those obtained using the app-based digital measurement system(myFitNote®). The results from the Bland & Altman analysis showed stable confidence intervals, confirming high agreement between the two methods. The absolute error rate was found to be 5.78%. Consequently, the validity of the TUG test between the expert method using a stopwatch and the AI-based physical function testing system (myFitNote®) was confirmed to be high.