TY - EJOU
AU - Yao, Jiepeng
AU - Peng, Zhanjia
AU - Liu, Jingjing
AU - Fan, Chengxiao
AU - Wang, Zhongyi
AU - Huang, Lan
TI - A Time-Varying Parameter Estimation Method for Physiological Models Based on Physical Information Neural Networks
T2 - Computer Modeling in Engineering \& Sciences
PY - 2023
VL - 137
IS - 3
SN - 1526-1506
AB - In the establishment of differential equations, the determination of time-varying parameters is a difficult problem,
especially for equations related to life activities. Thus, we propose a new framework named BioE-PINN based on a
physical information neural network that successfully obtains the time-varying parameters of differential equations.
In the proposed framework, the learnable factors and scale parameters are used to implement adaptive activation
functions, and hard constraints and loss function weights are skillfully added to the neural network output to speed
up the training convergence and improve the accuracy of physical information neural networks. In this paper, taking
the electrophysiological differential equation as an example, the characteristic parameters of ion channel and pump
kinetics are determined using BioE-PINN. The results demonstrate that the numerical solution of the differential
equation is calculated by the parameters predicted by BioE-PINN, the Root Mean Square Error (RMSE) is between
0.01 and 0.3, and the Pearson coefficient is above 0.87, which verifies the effectiveness and accuracy of BioE-PINN.
Moreover, real measured membrane potential data in animals and plants are employed to determine the parameters
of the electrophysiological equations, with RMSE 0.02-0.2 and Pearson coefficient above 0.85. In conclusion, this
framework can be applied not only for differential equation parameter determination of physiological processes
but also the prediction of time-varying parameters of equations in other fields.
KW - Physics-informed neural network; differential equation; bioelectrical signals; inverse problems
DO - 10.32604/cmes.2023.028101