Implementación de un prototipo de monitoreo de postura a nivel de cabeza, cuello, hombros y cintura escapular usando cámara de profundidad, caso de estudio de problemas músculo esqueléticos

Abstract

It is difficult for physiotherapists to know in depth the postural behavior of their patients. This is important information to have a complete understanding of the patient's context and the risk factors that contribute to the appearance of musculoskeletal disorders. In this context, we present the development and testing of a real time prototype with skeleton tracking of postures adopted by a person while sitting in front of a computer. The prototype uses a depth camera to obtain the spatial coordinates of the body's joints. These points are used to calculate the angles that describe the individual's posture. A data set of angles is built to train 3 classification methods RNN, RFC and GBC. For all 3 classification methods, the accuracy rate exceeds 70% for the 5 postures studied. The accuracy rate of the prototype used in a physiotherapy case study is not affected by more than 10% compared to the results obtained under controlled conditions. In addition, the results obtained in the study case assisted a specialist to develop a treatment plan.