<![CDATA[Smart wheelchairs are an innovation designed to facilitate user mobility while monitoring their condition in real time. One of the main features developed is an integrated weight reading system. However, the accuracy of the sensor is still affected by sitting posture, body position, and surrounding environmental conditions. This study aims to improve the accuracy of the weighing system on smart wheelchairs by applying linear regression analysis as a sensor calibration method. Data collection was conducted under four conditions of use, namely sitting upright, sitting tilted, walking while sitting upright, and walking while sitting tilted, which represent variations in user load distribution. The calibration model was constructed using the average sensor reading data and evaluated using the R², MAE, and MAPE parameters. The results showed a significant improvement in accuracy with an R² value of 1.0000, MAE of 0.0687 kg, and MAPE of 0.111%, as well as a decrease in the average error from ±1.2 kg to ±0.07 kg after the calibration process. The linear regression method proved to be effective in improving the accuracy of sensor readings with light computational calculations. This study also demonstrates the potential of linear regression as an efficient lightweight calibration method for IoT-based medical systems, particularly on devices such as ESP32 or Arduino that display real-time, high-precision body weight measurements.]]>
Copyrights © 2025
