pinn-neuronal-parameter-estimation

You can estimate neuronal model parameters from partial voltage observations by using Physics-Informed Neural Networks (PINNs) to fit observed data and infer unobserved biophysical parameters simultaneously.

State reconstruction for unobserved variables in neuron models works by using PINNs with automatic differentiation to train trainable biophysical parameters that fit partial voltage data and reconstruct hidden states.

Yes, you can use PINNs for parameter estimation in Morris-Lecar and fast-slow bursting neuron models, as the method is applicable to multi-scale neuronal systems and breathing neurons.

No, you do not need full voltage traces; PINNs enable robust parameter estimation and state reconstruction with limited data, relying on partial observations rather than complete datasets.

PINNs are robust against inaccurate initial parameter guesses because the physics-informed neural network approach is insensitive to initial values while optimizing trainable biophysical parameters against observed data.