| Statistics used in biological industry, which is different from that used in other area, focuses more on efficacy and safety analysis. Most of these bio-studies are for approval of drug or device from food and drug administration. Statistical analysis is the key role in the development of new drug. And precise, scientific and effective statistical analysis can be the last security gate for a drug’s real efficacy and safety, directly or indirectly affect the health of human beings. Therapeutic area of Diabetes is one of the most common area in recent years which is relative mature with many studies. Thus diabetes study can be a typical biological study to reflect the statistical analysis often used in real projects. In many other papers, more and more factors are explored to study the potential affective ones which contribute to a diabetes disease. Or, some paper may only focus on the efficacy analysis, or just adding more factors, these directions may block the real systematic analysis of a bio-study.This paper is trying to start from a type Ⅱ Diabetes mellitus study, using some common used analysis statistical models, including t-test, ANCOVA model, MMRM model, cLDA model, Logistical Regression model etc.. The covariate is only limited to BMI group and corresponding baseline value, which is armed to simplify and systematically describe the application of statistics in projects. This paper is also trying to discuss the difference between cLDA model and ANCOVA model and MMRM model. Start from theory, discussion from practices, and conclusion from comparison. |
