This paper describes a method for the estimation of bio-signals based on human motion in daily life for an integrated visualization system. The recent advancement of computers and measurement technology has facilitated the integrated visualization of bio-signals and human motion data. It is desirable to obtain a method to understand the activities of muscles based on human motion data and evaluate the change in physiological parameters according to human motion for visualization applications. We suppose that human motion is generated by the activities of muscles reflected from the brain to bio-signals such as electromyograms. This paper introduces a method for the estimation of bio-signals based on neural networks. This method can estimate the other physiological parameters based on the same procedure. The experimental results show the feasibility of the proposed method.