Using multivariate logistic regression to study the relationship between mechanical inputs and neural responses in mechanoreceptor neurons

An original statistical methodology is used to test whether the responses of mechanoreceptor neurons better encode tissue stress or strain during dynamic loading. Experiments were conducted by stretching samples of innervated tissues in vitro. The input function was a Pseudorandom Gaussian Noise (PGN) sequence. Stress and strain and their rates of change were measured and used as continuous input variables; neuron discharges were used as the binary outcome variable. The data were organized to fit a Multivariate Logistic Regression (MLR) model, and the strength of coupling between neural discharge and each mechanical variable was assessed using the Odds Ratios.