Identification of time-varying neural dynamics from spike train data using multiwavelet basis functions

Area of Science:

• Computational Neuroscience
• Systems Neuroscience
• Signal Processing

Background:

• Understanding neural dynamics is crucial for deciphering learning mechanisms in animals.
• Neural processes underlying learning are time-varying, posing significant modeling challenges.

Purpose of the Study:

• To develop a novel framework for modeling time-varying neural dynamics using spike train data.
• To accurately track rapid and slow parameter changes in neural systems.

Main Methods:

• A multiwavelet-based time-varying generalized Laguerre-Volterra (TVGLV) modeling framework was developed.
• Forward orthogonal regression (FOR) with mutual information (MI) was used for model selection.
• Generalized linear model fitting was employed for parameter estimation from spike train data.

Main Results:

• The proposed TVGLV approach successfully identified parameters in synthetic and real retinal spike train data.
• The method demonstrated high sensitivity and accuracy in tracking both gradual and abrupt parameter changes.
• Performance was robust without requiring prior knowledge of spike train characteristics.

Conclusions:

• The multiwavelet-based TVGLV framework offers a robust computational tool for analyzing nonstationary neural properties.
• This approach can track general forms of time-varying neural dynamics.
• Potential applications include investigating spatio-temporal information in biomedical spiking signals.