Leg fatigue can influence the gait patterns, therefore declining the postural stability and the motor performance, increasing the risk of falls. In order to improve the earlier detection of risks and the application of fall prevention strategies, automated solutions based on gait analysis must be developed. A sector of the population at risk is the workforce where a majority of workers admits to be fatigued and where falls can lead to serious workplace injuries or even deaths. In these cases, having the ability to detect if the user is fatigued in real time by simply using the motion sensors on the smartphone and processing it with machine learning can lead to the prevention of falls and the consequences these bring.
Phones and wearable devices can be used to extract inertial sensor’s data to provide enough information for the fatigue detection. Supervised machine learning algorithms, such as Support Vector Machines (SVM) and Neural Networks, will be used to process this information for fatigue level classification. Their performance will then be compared to find the best algorithm for fatigue detection. In addition to this comparative work, different conditions for the data collection and processing were tested in an effort to discover the optimal conditions for the implementation of the algorithms.
Author: André Campos
Type: MSc thesis
Partner: Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa