<![CDATA[Height measurement is an important clinical indicator for nutritional status assessment but is difficult to perform on patients with mobility limitations or spinal disorders. This study validates the accuracy of the Chumlea Algorithm in predicting height using automatic knee height measurement based on ultrasonic sensors on the SMATSI Smart Wheelchair. A cross-sectional validation study was conducted on 60 measurement data (30 males and 30 females) from an elderly adult population. The best algorithm selection used Multi-Criteria Decision Making (MCDM) methods: Simple Additive Weighting (SAW) and Analytical Hierarchy Process (AHP) with four criteria (Bias, MAE, RMSE, Correlation). Results show that the Chumlea Algorithm has the highest accuracy with an average difference of 0.3cm (males) and 0.8 cm (females) from actual height, not statistically significant (p > 0.05). MCDM analysis confirmed Chumlea as the best choice with first ranking in both methods (SAW: 1.000; AHP: 0.524). Pearson correlation shows very strong relationship (r = 0.82-0.85, p < 0.001). Bland-Altman analysis shows Limits of Agreement (LoA) -4.0 to +5.1 cm, which is within clinically acceptable range for BMI calculation in this population. This accuracy is much better than conventional studies in Indonesia showing bias of 3.44–7.33 cm. Automatic knee height measurement using ultrasonic sensors eliminates systematic errors of manual measurement.]]>
Copyrights © 2026
