Comparing parameter-reduction methods on a biophysical model of an auditory hair cell

• DOI: 10.1103/physrevresearch.6.033121
• Publication Date: 2024-07-31T14:11:29Z
• Authors: Joseph M. Marcinik; Martín A. Toderi; Dolores Bozovic
• Keywords: Akaike information criterion; Overfitting; Bayesian information criterion

Biophysical models describing complex cellular phenomena typically include systems of nonlinear differential equations with many free parameters. While experimental measurements can fix some parameters, those internal processes frequently remain inaccessible. Hence, a proliferation parameters risks overfitting the data, limiting model's predictive power. In this study, we develop systematic methods, applying statistical analysis and dynamical-systems theory, to reduce parameter count in biophysical model. We demonstrate our techniques on five-variable computational model designed describe active, mechanical motility auditory hair cells. Specifically, use two measures, total-effect PAWN indices, rank each by its influence selected, core properties With resulting ranking, most less influential yielding five-parameter refined validate theoretical recordings active hair-bundle motility, specifically using Akaike Bayesian information criteria after obtaining maximum-likelihood fits. As result, determine system's which illuminate key elements cell's overall features. Even though concrete example, provide general framework, applicable other systems. Published American Physical Society 2024