Time series model with the autoregressive structure on clinical data

Document Type : Original Scientific Paper

Authors

Department of Statistics, Marvdasht Branch, Islamic Azad University, Marvdasht, Iran

Abstract

This research explores a time series model with an autoregressive structure applied to clinical data‎. ‎The motivation for selecting this topic is to develop a probabilistic model for nonlinear time series with interventions in clinical datasets‎. ‎By distinguishing between the phases before and after the intervention and analyzing the changes during the intervention‎, ‎the analysis yields a precisely estimated effect of the intervention‎. ‎The iterative scheme expectation/conditional maximisationeither algorithm is proposed for parameter estimation‎, ‎and the observed information matrix is derived analytically‎. ‎A key focus of data analysis is evaluating the robustness of the model's estimates and understanding how minor local disturbances influence the model‎. ‎The local impact of the model is thoroughly analyzed across three disturbance scenarios‎. ‎To assess the performance of the proposed methods‎, ‎simulated datasets are presented‎, ‎incorporating expectation/conditional maximisationeither estimates to demonstrate the robustness of estimates in the presence of influential outliers‎. ‎Finally‎, ‎the proposed method is successfully applied to model new COVID-19 time series cases in the Czech Republic‎. ‎Appropriate criteria confirm the applicability of the proposed process‎, ‎alongside the impact of diagnostic analysis.

Keywords

Main Subjects