Predicting Stride Length from Acceleration Signals Using Lightweight Machine Learning Algorithms

In the field of motion analysis and behavioral movement research, accurately measuring and predicting stride length from acceleration signals plays a critical role. Recent studies have undertaken the development and evaluation of numerous machine learning models to address this issue, focusing on the deployment of lightweight and efficient models on devices with limited computational capabilities, such as microcontrollers. Among the algorithms examined, random forest, k-nearest neighbors, support vector machines, decision trees, and extreme gradient boosting have been selected due to their balance between prediction accuracy and computational lightness. These algorithms not only ensure a high degree of accuracy in predicting stride length from collected acceleration data but also ensure that the models can be deployed directly on smart wearable devices and microcontrollers, extending the practical application of motion monitoring systems. The outcomes of these machine learning models promise to contribute to the development of personalized health monitoring solutions, as well as in applications related to sports and functional rehabilitation, where analyzing and improving gait is extremely important.