Predict human body indentation lying on a spring mattress using a neural network approach

Abstract

This article presents a method to predict and assess the interaction between a human body and a spring mattress. A three-layer artificial neural network model was developed to simulate and predict an indentation curve of human spine, characterized with the depth of lumbar lordosis and four inclination angles: cervicothoracic, thoracolumbar, lumbosacral and the back-hip (β). By comparing the spinal indentation curves described by the optimal evaluation parameters (depth of lumbar lordosis, cervicothoracic, thoracolumbar and lumbosacral), a better design of five-zone spring mattresses was obtained for individuals to have an effective support to the main part of the body. Using such approach, an operating process was further introduced, in which appropriate stiffness proportions were proposed to design mattress for the normal body types of Chinese young women. Finally, case studies were undertaken, which show that the method developed is feasible and practical.

Keywords: Indentation curve; artificial neural network; five-zone mattress; simulation; spine.