Related Experiment Video
Updated: Sep 23, 2025
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Clinical prediction rule for locomotive syndrome in older adults: A decision tree model
Shigeharu Tanaka1, Hungu Jung2, Shunsuke Yamashina2
1School of Rehabilitation, Kanagawa University of Human Services, Yokosuka, Kanagawa, Japan; Graduate School of Humanities and Social Sciences, Hiroshima University, Higashihiroshima, Hiroshima, Japan.
Background:
No previous studies have proposed a clinical prediction rule that analyzes the factors related to the severity of locomotive syndrome. This study developed and assessed a clinical prediction rule for the severity of locomotive syndrome in older adults.
Methods:
A total of 186 patients were assessed using the locomotive syndrome risk test. Classification and regression tree methodologies were used to develop the clinical prediction rule. This study developed three prediction models based on the severity of the locomotive syndrome, of which Model 3 assessed the most severe condition. The following potential predictive factors were measured and entered into each model; single-leg standing time, grip strength, preferred and maximum walking time, and timed up and go test.
Results:
The single-leg standing test (≤59.4 or >59.4 s) was the best single discriminator for Model 1. Among those with a single-leg standing time >59.4 s, the next best predictor was grip strength (≤37.8 or >37.8 kg). In Model 2, the single-leg standing test was also the best single discriminator (≤12.6 or >12.6 s). Among those with a single-leg standing time ≤12.6, the next best predictor was TUG (≤7.9 or >7.9 s). Additionally, among those with a single-leg standing time >12.6, the next best predictor was single-leg standing time (≤55.3 or >55.3 s). In Model 3, predictive value in Model 2 was the best single discriminator (0 or 1). Among those with 1, the next best predictor was maximum walking time (≤3.75 or >3.75 s). The area under the receiver operating characteristic curves of Models 1, 2, and 3 were 0.737, 0.763, and 0.704, respectively.
Conclusions:
A clinical prediction rule was developed to assess the accuracy of the models. These results can be used to screen older adults for suspected locomotive syndrome.
