Physics-informed neural networks for inverse identification of α1-adrenergic receptor non-desensitization mechanisms
Mathematical Biosciences · 2027-01 · 403 · 109846
DOI: 10.1016/j.mbs.2026.109846 ↗Kurz gefasst
Fachartikel in der Zeitschrift Mathematical Biosciences, 2027-01. Verfasst von Shengying Zhou und 1 weiteren Autor:innen. Beteiligte Einrichtung: College of Mathematics, Taiyuan University of Technology, China. Die Arbeit nennt 34 Literaturangaben.
Abstract (Originalsprache)
This study investigates the aberrant activation and "non-desensitization" phenomenon of α1-adrenergic receptors (α1-AR) in hypertension. To systematically elucidate the mechanisms underlying receptor-mediated vascular smooth muscle contraction, a unified kinetic model was constructed, comprising a control group, an endocytic dysfunction group, a conformational change group, and a composite model integrating multiple regulatory processes. The model explicitly accounts for receptor binding, phosphorylation, endocytosis, and conformational transitions. Using rat vascular ring tension measurements, a quantitative mapping between model states and experimentally observed contractile responses was established. A physics-informed neural network (PINN) implementation combining a predefined linear warm-up schedule for the physics-loss weight with sequential Adam and L-BFGS optimization was employed to infer key kinetic parameters while balancing data fitting and kinetic constraints during training. Numerical comparisons based on fitting errors and AIC indicated that the composite model provided the most favorable balance between data agreement and model complexity among the candidate mechanisms. Analyses revealed that α1-AA sustains receptor activation and continuously amplifies downstream signaling by enhancing binding affinity, stabilizing the active conformation, and suppressing endocytosis, thereby promoting persistent vasoconstriction. These findings provide a quantitative explanation for the "non-desensitization" of α1-AR and demonstrates the practical utility of PINNs for parameter estimation and mechanistic analysis in complex biological systems.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- College of Mathematics, Taiyuan University of Technology, China.
- College of Mathematics, Taiyuan University of Technology, China. Electronic address: xywang0708@126.com.
Zitieren
APA
Zhou, S., & Wang, X. (2027). Physics-informed neural networks for inverse identification of α1-adrenergic receptor non-desensitization mechanisms. Mathematical Biosciences, 403, 109846. https://doi.org/10.1016/j.mbs.2026.109846
Vancouver
Zhou S, Wang X. Physics-informed neural networks for inverse identification of α1-adrenergic receptor non-desensitization mechanisms. Mathematical Biosciences. 2027;403:109846. doi:10.1016/j.mbs.2026.109846
BibTeX
@article{Zhou2027,
author = {Zhou, Shengying and Wang, Xiaoyun},
title = {Physics-informed neural networks for inverse identification of α1-adrenergic receptor non-desensitization mechanisms},
journal = {Mathematical Biosciences},
year = {2027},
volume = {403},
pages = {109846},
doi = {10.1016/j.mbs.2026.109846}
}
EUVIMED