Dynamical physically structured data modeling vs. classical time series analysis: A case study related to clinical trial data analysis

Abstract

At the time series analysis’s core is the fitting of a preselected model—in the most straightforward case a linear combination of basis functions—to experimental data with one of the performance indices, for example, least squares criterion. However, no matter what the type of such a model is and whatever be the fitting performance index, such an approach does not stipulate for reliance on the understanding those dynamical laws of physics dictating the observable data behavior, given their complexity or uncertainty. Our research looks into a question: “What benefits or advantages could mathematical modeling of such laws be capable of giving when included in nature or experimental data analysis?—nowhere more so than in deciding on life-changing events based on time series analysis in medical practice”.