XJTU makes new progress in functional data analysis to tackle COVID-19

2021-12-06  []


New research results on functional data analysis from Dr Liu Hua of the School of Economics and Finance of Xi'an Jiaotong University (XJTU) and collaborators have been published in the world's top journal in statistical science.  


Based on the characteristics of a large amount of new functional data, a new type of dynamic interactive semiparametric function-on-scalar (DISeF) model was proposed.  





Dr Liu worked with Dr You Jinhong of the Shanghai University of Finance and Economics and Professor Jiguo Cao of Simon Fraser University to explore the major impact of the COVID-19 epidemic on countries around the world.  


The researchers used the DISeF model and corresponding statistical inference methods to analyze COVID-19 cases and MRIEEG patterns.  


For example, in the COVID-19 case analysis, they found that factors such as the aging of the population, social economy and medical resources including the number of hospital beds per thousand people and the number of medical staff have an interactive effect on the mortality rate, which varies at different epidemic stages.  


Based on the index parameter vector, researchers have also developed a health infrastructure index system that equates to the COVID-19 mortality rate in 141 countries around the world, which can help the World Health Organization guide countries to effectively fight the epidemic.  


By combining tensor B-spline approximation techniques, a three-step effective estimation method was proposed to estimate the unknown two-dimensional function, index parameter vector and high-dimensional covariance matrix in the model, establishing the asymptotic properties of these estimators including their convergence rate and asymptotic distribution.  


In addition, a test method was proposed to judge whether the dynamic interaction in the model changes with time or space, and the asymptotic normality of the test statistic was given. The researchers found that there are interactions in these cases that vary with indicator parameters of vector, time or space.  


Entitled ADynamic Interaction Semiparametric Function-on-Scalar Model, the research results were published on the Journal of the American Statistical Association, a premier journal of statistical science with Liu as the first author.  


Link to the paper:https://www.tandfonline.com/doi/full/10.1080/01621459.2021.1933496