| ABSTRACT | 第5-6页 |
| 中文摘要 | 第7-9页 |
| ABBREVIATION | 第9-10页 |
| Background | 第10-12页 |
| Methods | 第12-22页 |
| Study Design | 第12-13页 |
| Definitions of NAFLD and MetS | 第13-14页 |
| Missing Data Imputation and Variable Assignments | 第14-15页 |
| Statistical Analysis | 第15-22页 |
| Generalized Estimating Equation(GEE) | 第16-17页 |
| Causal Inference Network---Bayesian Network | 第17-20页 |
| Causal Effect Identification | 第20-21页 |
| Rank the relative importance of pathways in pathogenesis | 第21-22页 |
| Results | 第22-50页 |
| Characteristics of sub-cohorts | 第22-24页 |
| Potential bi-directional causality between NAFLD and MetS by GEE | 第24-41页 |
| Potential causality from NAFLD to MetS(sub-cohortA) | 第24-34页 |
| Potential causality from MetS to NAFLD(sub-cohortB) | 第34-41页 |
| Reciprocal causal inference by Bayesian Network | 第41-50页 |
| Causal inference from NAFLD to MetS in sub-cohortA | 第41-45页 |
| Causal inference from MetS to NAFLD in sub-cohortB | 第45-50页 |
| Disussion | 第50-54页 |
| Potential bidirectional causality between NAFLD and MetS | 第50页 |
| Reciprocal causality between NAFLD and MetS | 第50-52页 |
| New findings from Bayesian network | 第52-53页 |
| About statistical methods | 第53-54页 |
| Limitation | 第54页 |
| Conclusion | 第54-55页 |
| References | 第55-61页 |
| Practice report | 第61-71页 |
| Acknowledgement | 第71-72页 |
| Publications | 第72-73页 |
| 附表 | 第73页 |
