Efficient parameter inference in networked dynamical systems via steady states: A surrogate objective function...

Area of Science:

• Networked dynamical systems
• Computational biology
• Systems ecology

Background:

• Parameter inference is vital for predicting nodal dynamics in networked systems.
• Traditional methods often require time-series data, which is not always available.
• Noisy steady-state data presents a challenge for parameter estimation.

Purpose of the Study:

• To develop an efficient method for inferring dynamical parameters from noisy steady-state data.
• To reduce the computational complexity associated with traditional parameter estimation techniques.
• To improve the accuracy of predicting future equilibria in networked systems.

Main Methods:

• Introduced a surrogate objective function using decoupled equations to compute steady states.
• Optimized the surrogate objective function to approximate ground truth steady states.
• Reduced computational demand by avoiding repeated simulations of coupled ordinary differential equations.

Main Results:

• Achieved more accurate steady-state approximations compared to noisy observations.
• Successfully predicted future equilibria following network topology changes.
• Demonstrated effectiveness across ecological, gene regulatory, and epidemic network models.

Conclusions:

• The proposed method offers an efficient and effective approach for parameter estimation from steady-state data.
• This technique has the potential to enhance predictions in various networked dynamical systems.
• Enables robust parameter inference even with limited or noisy observational data.