Causal Recommender Systems

Overview

Biography

Xiao-Hua (Andrew) Zhou, PKU Distinguished  Chair Professor and  Chair of the Department of Biostatistics, Vice Dean of National Institute for Regulatory Science of Drug and Medical Device at Peking University, and Vice Dean of Peking University Chongqing Research Institute for Big Data, Fellow of the American Association for the Advancement of Science (AAAS), Fellow of the American Statistical Association (ASA), Fellow of the Institute of Mathematical Statistics (IMS).

Professor Zhou has published more than 280 SCI academic papers in top international statistical and biostatistical journals such as J. R. Statist. Soc. B, Journal of the American Statistical Association, Annals of Statistics, Biometrika,  Biometrics, etc., of which more than 170 are the first or corresponding authors. Professor Zhou's research focuses on statistical methods in diagnostic medicine, causal inference in biostatistics and and machine learning, the analysis of skewed data, mathematical and statistical modeling of the occurrence and development of major epidemics, and statistical methods for Chinese medicine.  He served as a member of the US Federal Government Food and Drug Administration (FDA) Medical Devices and Radiological Health Advisory Committee. He has won the Research Career Scientist Award from the U.S. Federal Government Department of Veterans Affairs, Distinguished Overseas Young Scientist Award of the Chinese National Natural Science Foundation, the China Ministry of Education Oversee Distinguished Culture and Education Expert, and the China Ministry of Education Oversee Distinguished Teacher, China  National Overseas High-level Talent Program Expert, the Mitchell Prize of the International Society for Bayesian Analysis, the Best Research Paper Award selected by SCIENCE CHINA-Mathematics. And He has been principle Investigator of major research projects of  the National Natural Science Foundation of China and the key research and development projects of the Ministry of Science and Technology.

Statistical Methods in Diagnostic Medicine

Prof. Zhou has made important and original contributions and developed a large number of influential statistical methods. His most extensive and comprehensive contribution to the accuracy assessment of diagnostic tests has been the development of methods to address multiple types of missing data, including the critical problem of Verification and Imperfect Gold Standard Bias. These methods provide medical researchers with important tools to better design studies and analyze relevant data. As a result of his vital in the field, Professor Zhou published the book, entitled “Statistical Methods in Diagnostic Medicine”, the first comprehensive statistical textbook in diagnostic medicine. The first edition of the book was published by Wiley & Sons in 2002 and the second edition in 2011, and it becomes a standard textbook of statistics in diagnostic medicine, and has been cited 2,620 times.

Causal Inference

Professor Zhou has made important contributions in the causal inference in the research on Identification of Causal Effect, Robustness of Estimators, Causal Inference with Truncation-by-Death, Causal Inference of Randomized Encouragement Designs, Precision Medicine. Professor Zhou proposed to use instrumental variable method to combine the intended treatment effect with the real treatment effect, using Bayesian method for inference and sensitivity analysis, put forward a series of new methods and theories of correlation causal effect estimation. Due to the importance of the method, Zhou and his collaborators received the Mitchell Prize from the International Society for Bayesian Analysis, and the paper was selected for discussion by Biostatistics and Biometrics.  In the area of causal precision medicine, Professor Zhou first proposed the use of Biomarker Adjusted Treatment Effect (BATE) curve and Covariate-Specific Treatment Effect (CSTE) curve to represent the conditional average treatment effect under a given biomarker level,and it provides a uniform inferential tool in making individualized treatment decisions, and strictly proved the mathematical properties of the proposed new statistical method. The research results won the Outstanding Paper Award of the Chinese Journal of Science and Mathematics in 2015. Then, Professor Zhou further proposed the method of CSTE curve with binary outcome variable and Confidence Band, and extended the CSTE curve to high-dimensional covariable scenarios, and optimaized its mathematical theory.

Chinese Medicine Statistical Methods

For a long time, due to the lack of efficiency evaluation tools, it is difficult to evaluate the effectiveness of traditional Chinese medicine, resulting in a weak position in the competition with western medicine. Professor Zhou is interested in developing statistical and machine learning  methods in  traditional Chinese medicine. Based on the diagnostic and causal theoretical research, he provides support for traditional Chinese medicine in terms of statistical methodology, makes the evaluation of TCM efficacy of TCM syndrome and diagnosis scale more accurate and reasonable, plays a positive role in promoting the normalization and standardization of TCM research, and solves the difficulties faced by traditional Chinese medicine to a large extent. Also, it is conducive to improving the scientificity and credibility of TCM related research. In 2021, the core technology of his research and development, "Evaluation of syndrome diagnosis tools and statistical method innovation of efficiency effect causal inference", won the second prize of Science and Technology Progress Award of the World United TCM International Contribution Award. In addition, Professor Zhou led a team in Peking University Chongqing Research Institute of Big Data to apply for three software Copyrights, and on this basis developed the first TCM syndrome diagnosis and efficacy evaluation platform based on big data in China, enabling grassroots TCM, promoting TCM service upgrading, and better serving patients.