A Framework for Parameter Estimation and Uncertainty Quantification in Systems Biology Using Quantile Regression and Physics-Informed Neural Networks
by Haoran Hu, Qianru Cheng, Shuli Guo, Huifang Wen, Jing Zhang, Yongqi Song, Kaiqun Wang, Di Huang, Hui Zhang, Chaofeng Zhang, and Yanhun Shan
Read the paper
This study introduced a novel framework integrating quantile regression with Physics-Informed Neural Networks to enhance parameter estimation and uncertainty quantification in systems biology models. The approach demonstrated superior accuracy in parameter estimation, stronger correlation between uncertainty and noise levels, and moderate computational costs. By efficiently addressing data limitations and noise, this framework offers a powerful tool for advancing predictive biological modeling, with potential applications in drug development, disease modeling, and understanding complex biological interactions. Its scalability and reliability position is as a promising solution for real-world systems biology challenges.

The illustration of the PINN-based quantile regresion approach.