Muhammad, Naseer (2026) Mathematical Modeling of Neuronal Dynamics And Population Heterogeneity For Large-Scale Neuronal Networks. [Tesi di dottorato]

Documento PDF Thesis_Muhammad_Naseer.pdf Visibile a [TBR] Utenti registrati Download (11MB) | Richiedi una copia

Abstract

This thesis develops and evaluates a data-driven modeling framework for reproducing the firing behavior of hippocampal CA1 pyramidal neurons using the Adaptive Generalized Leaky Integrate-and-Fire (A-GLIF) model. The framework was applied to whole-cell patch-clamp recordings from 109 neurons obtained from GFP-expressing mice, wild-type Control mice, WT-Volkmar mice, and APPPS1 transgenic mice. A-GLIF parameters were individually fitted for each neuron, and model performance was validated by comparing experimental and simulated spike trains across multiple stimulation currents. Mann–Whitney U-tests revealed no statistically significant differences in spike timing (p > 0.05), indicating strong agreement between recorded and simulated activity. Model accuracy was further quantified using the Spike Train Similarity Measure (STSimM), which consistently demonstrated high temporal accuracy (≥ 70%) and F-scores (≥ 60%). The fitted models were subsequently used to reconstruct missing stimulation ranges, providing initial estimates of permissible firing boundaries and spike-time variability. Based on these estimates, heterogeneous neuron populations for all groups were generated by perturbing 12 key A-GLIF parameters. To ensure biological plausibility, only neuron copies whose firing patterns remained within experimentally observed variability envelopes were retained. Overall, this work presents a computationally efficient and biologically grounded approach for generating realistic neuronal populations, enabling scalable and physiologically consistent large-scale simulations, while also providing insight into excitability alterations associated with Alzheimer’s disease.

Downloads

Actions (login required)

Modifica documento