Bayesian Estimation of Phase Dynamics Based on Partially Sampled Spikes Generated by Realistic Model Neurons

Area of Science:

• Computational Neuroscience
• Systems Neuroscience
• Data Analysis

Background:

• Stable rhythmic activity in dynamic systems can be modeled using phase oscillators.
• A Bayesian approach was previously developed to extract phase dynamics from time-series data.
• Spike data offers limited phase information compared to continuous time-series data.

Purpose of the Study:

• To extend a Bayesian method for extracting phase dynamics from spike data.
• To evaluate the method's performance on simulated neuronal network spike data.
• To assess the method's ability to infer neural interaction functions and synaptic connections.

Main Methods:

• Application of a Bayesian inference method to simulated spike trains from a neuronal network model.
• Comparison of estimated phase dynamics with theoretically derived dynamics.
• Analysis of the inferred interaction function to identify synaptic connections.

Main Results:

• The extended Bayesian method successfully extracted modeled phase dynamics from spike data.
• Accurate extraction of the interaction function was achieved with sufficient data.
• The method demonstrated the ability to infer synaptic connections based on the estimated interaction function.

Conclusions:

• The developed Bayesian method is applicable to spike data for analyzing rhythmic neural systems.
• This approach provides a practical tool for understanding the dynamic properties of neural networks.
• The findings support the utility of phase oscillator dynamics in computational neuroscience research.