Multimodal ARMAX Model for Characterization of Human Body System

Viewing the holistic human body as a linear time-invariant (LTI) system and fusing multimodal physiological signals to understand the physical state by borrowing the idea from systems medicine (SM), we propose a multimodal autoregressive moving average with exogenous terms (MARMAX) model to characterize a human body. The model parameter matrices are calculated from the real world multimodal physiological signals using two stage least square (2SLS) method for personalization. Based on the personalized MARMAX model, one can predict the physiological signals of a certain person. Taking wrist pulse signals and tongue images as output signals for LTI human systems, experimental results are provided to evaluate the effectiveness of the proposed model.