Ligong Chen

Education and Working Experience
Ligong Chen (in English naming convention) or Chen Ligong or Chen Li-Gong (in Chinese pronunciation convention). Chinese name: . He was born on February 24th, 1964 in the small town of Fengkou, Honghu City in Hubei province, China. His father is a retired geography teacher in the Honghu Second High School in Fengkou. After graduating from the former Tongji Medical University and obtaining a degree of medical doctor in 1987, Ligong Chen was employed by the Department of Preventive Medicine/Epidemiology and Health Statistics of the School of Public Health of the university, and his position's title was from teaching assistant in the beginning to associate professor in 2000. From September of 1994 to June of 1997, he was a student of Professor Songlin Yu, a famous expert in Biostatistics in China, and did a research project on an economical evaluation for schistosomiasis control in the lake areas of China for his master degree of public health. He finally left the department and emmigrated to the United States of America in early 2002. Now he is an employee of the Henry M. Jackson Foundation (HJF) and is working in the Department of Preventive Medicine and Biometrics of the Uniformed Services University of the Health Science, (USUHS).
Research Career
From February of 1998, he began to study a new statistical method named functionalized general trichotomic regression analysis, or functionalized general trichotomous regression analysis, FGTRA.
In September of 1999, he published a philosophical paper about a progressive structure of intelligence on the Chinese journal of (Medicine and Philosophy), in which he elaborated an almost complete mechanism of the human’s brain working in an innovative scientific exploration.
In August of 2000 with the financial support of the Department of Education of China, he participated the Joint Statistical Meetings (JSM 2000) in Indianapolis, Indiana in the United States, and gave an oral presentation at the Section of General Methodology on a method of “functionalized critical regressive segmentation” for mutation models.
In December of 2000, he published a paper on the Chinese Journal of Public Health on how to seek an optimal threshold on an investment curve in order to amend the investment approach in schistosomiasis control. Although there is no significant innovation in the field of statistical methodology within this paper, he proposed a new strategy by introducing a single full-range regression analysis or fullwise regression analysis (FRA) into the classical piecewise regression analysis (PRA) and constructing a coefficient of residuals resistance (CRR). In this paper, he still took an optimization by using a maximal CRR to determine the optimal threshold as well as a set of optimal segmented models or piecewise models. A significant difference between his method and the classical PRA is that he can use the maximal CRR to evaluate his optimal piecewise models. Another significant difference is that he didn't take the enforced continuity as a necessary assumption, as the classical PRA did, but a real measured random point to estimate the optimal threshold.
He made a fresh start on the FGTRA in August of 2006 after he was employed by the HJF and worked in the Center for Prostate Diseases Research (CPDR), USUHS. In July of 2007 in Salt Lake City, Utah, he participated the JSM 2007 and gave an oral presentation at the Section of Statistical Computing on the FGTRA. At this time, He proposed a complete new analytical logic and a basic mathematical algorithm for doing a functionalized general trichotomy in regression analysis by symmetrically or asymmetrical segmeting a continuous random linear sample space into three threshold spaces: low, middle and high. In this innovative method, each unknown threshold will be considered as a random threshold variable (RTV). Therefore, he thoroughly gave up the optimization and reconstructed the coefficient of residuals resistance into a convergence rate of residuals (CRR) as a random weight variable so that he could use the random threshold variable and the random weight variable to estimate the unknown threhsold in a weighted mean and a weighted confidence interval (WCI). By defining a general weighted mean and a general weighted standard deviation (WSD) and proving two properties of sample size n in the general weighted mean and the WSD, he suggested to use a sum of weights to generalize all regular statistics thus a weighted standard error (WSE) can and only can be defined on the sum of weights so that he could take the WSE to estimate any random threshold in a weighted confidence interval. Based on criticising the enforced continuity assumption in the classical PRA and the weighted mean threshold, he tried to construct a continuity test for the piecewise models. The paper was collected in the Proceedings of the JSM 2007. However, there were some unconventional mathematical expressions in the paper, and were still some problems that were not well solved, especially about the optimization in the classical PRA. In his opinion, it is theoretically incorrect in the classical PRA. He must prove this opinion.
In August of 2009 in Washington DC, he participated the JSM 2009 and tried to define a new concept of random correspondence from the angle of the Probability Theory, and suggested that the constant or constant expectation (in Statistics) should be included in the random system as a starting point and an ending point, just as the zero is a starting point in the number system. At this conference, he systematically elaborated nine properties of random variable and seven axiomatic statements in Statistics. Based on these fundamentals, he tried to prove that the classical piecewise regression analysis and its modern successor Spline techniques based on the optimization and the enforced continuity assumption are theoretically incorrect. The method for improving the classical methodology is to introduce a weighted random measurement to estimate any random threshold in a weighted expectation and a weighted confidence interval thus to determine the statistical expectation of the random piecewise models, and build a continuity test for the piecewise models. In this new paper, he amended the unconventional mathematical exxpressions and further improved his own methodology by proposing an approach to adjust the symmetrical WCI into asymmetrical one for each random threshold and adding a new algorithm in the continuity test.
Reference
External Link
*http://www.meetingproceedings.us/2009/jsm/contents/papers/303243.pdf

Main Menu

See Also

Main Menu