<![CDATA[Gait analysis is essential for diagnosing movement disorders and monitoring rehabilitation progress; conventional methods are often costly and complex. This study aims to differentiate gait characteristics between individuals with obesity and those with normal body composition using a multimodal approach that integrates Inertial Measurement Unit (IMU) and electromyography (EMG) sensors. Data were collected from ten male participants (five classified as obese and five with normal body composition). IMU sensors were used to measure acceleration, angular velocity, and step count, while EMG sensors recorded muscle activity from the tibialis anterior and gastrocnemius muscles. We developed a real-time acquisition using ESP32 microcontrollers and Bluetooth Low Energy (BLE), and gait phase classification was performed using the Hidden Markov Model (HMM). Using heel-mounted sensors, the average step detection error ranged from 2.5% to 3.6%. IMU signals from obese participants indicated a shift in dominant gait phase from Initial Contact during slow walking to Loading Response during fast walking, with relative errors up to 27%. In contrast, participants with normal body composition exhibited more diverse and accurate phase distributions. EMG-based analysis provided more precise segmentation (with error rates as low as 0.47%). It revealed distinct muscle activation patterns: gastrocnemius activity was dominant during the Midswing or Midstance phases, while tibialis anterior activity peaked during Initial Contact, Initial Swing, or Loading Response. These findings suggest body composition significantly affects gait stability, phase transitions, and muscle activation patterns. Future work should explore advanced machine learning algorithms such as Long Short-Term Memory (LSTM) or Convolutional Neural Networks (CNN), integrate pressure sensors, and validate the system in real-world environments to enhance accuracy and reliability.]]>
Copyrights © 2025
