Related Experiment Video
Updated: Dec 14, 2025
10:50
Computational Modeling of Retinal Neurons for Visual Prosthesis Research - Fundamental Approaches
Published on: June 21, 2022
2.1K
Estimation of neuron parameters from imperfect observations
Joseph D Taylor1, Samuel Winnall1, Alain Nogaret1
1Department of Physics, University of Bath, Bath, United Kingdom.
Plos Computational Biology
|July 17, 2020
Summary
This study introduces a novel regularization method to accurately estimate neuron electrical properties from noisy data. The technique significantly enhances the probability of identifying optimal ion channel parameters, improving our understanding of neural circuits.
Area of Science:
- Computational Neuroscience
- Biophysics
- Systems Biology
Background:
- Estimating parameters for biological neuron electrical properties is crucial for understanding ion channel complements and neural circuit function.
- Synchronizing conductance models with membrane voltage time series can predict neuronal dynamics, but identifying true ion channel parameters is challenging.
Purpose of the Study:
- To present a regularization method that enhances convergence to optimal solutions for parameter estimation in biological neurons, especially with noisy data and unknown models.
- To address the theoretical challenge of identifying the exact set of biological ion channel parameters.
Main Methods:
- Developed a regularization method leveraging an offset in parameter space, arising from model nonlinearity and experimental error.
- Induced saddle-node bifurcations to transition from sub-optimal to optimal solutions by tuning the parameter space offset.
- Implemented adaptive sampling and stimulation protocols to reduce parameter correlations and meet identifiability requirements.
Main Results:
- The regularization method increased the probability of finding optimal ion channel parameters from 67% to 94.3%.
- Adaptive protocols successfully reduced parameter correlations, enhancing model identifiability.
- Demonstrated that optimal model parameters can be inferred from imperfect observations under specific observability and identifiability conditions.
Conclusions:
- The presented regularization method offers a robust approach for parameter estimation in computational neuroscience.
- Fulfilling observability and identifiability conditions is key to inferring accurate neuronal model parameters from experimental data.
- This work advances the ability to model and predict neuronal dynamics using realistic biological parameters.
