Modeling a Nonlinear Biophysical Trend Followed by Long-Memory Equilibrium with Unknown Change Point

Abstract

Measurements of many biological processes are characterized by an initial growth period followed by an equilibrium period. Scientists may wish to quantify features of the growth and equilibrium periods, as well as the timing of the change point between the two. Specifically, we are motivated by problems in the study of electrical cell-substrate impedance sensing (ECIS) data. ECIS is a popular new technology which measures cell behavior noninvasively. Previous studies using ECIS data have found that different cell types can be classified by their equilibrium behavior. Likewise, it has been hypothesized that the presence of certain infections can be identified from cells' equilibrium behavior. However, it can be challenging to identify when equilibrium has been reached, and to quantify the relevant features of cells' equilibrium behavior. In this paper, we assume that measurements during the growth period are independent deviations from a smooth nonlinear function of time, and that measurements during the equilibrium period are characterized by a simple long memory model. We propose a method to simultaneously estimate the parameters of the growth and equilibrium processes and locate the change point between the two. We find that this method performs well in simulations and in practice. When applied to ECIS data, it produces estimates of change points and measures of cell equilibrium behavior which offer improved classification of infected and uninfected cells.

Likelihood Inference for Possibly Non-Stationary Processes via Adaptive Overdifferencing

“…Last, we compute BFR( = 5 × 10 −16 ) exact likelihood estimates of the memory parameter for several observed time series of electric cell-substrate impedance sensing (ECIS) measurements, which is a relatively new tool for monitoring the growth and behavior of cells in culture (Giaever and Keese, 1991). Previous research has suggested that ARFIMA(0, d, 0) models may be appropriate for ECIS measurements, and it has been hypothesized that the memory parameter d may vary by cell type and contamination status (Giaever and Keese, 1991;Tarantola et al, 2010;Gelsinger et al, 2020;Zhang et al, 2020). We consider ECIS measurements from eight experiments, four conducted using Madin-Darby canine kidney (MDCK) cells and African green monkey kidney epithelial (BSC-1) cells.…”