中华儿科杂志
2024年 · 第62卷第05期
中华儿科杂志
- 全部
- 述评
- 标准·方案·指南
- 指南解读
- 儿童保健相关研究
- 临床研究与实践
- 病例报告
- 综述
- 继续教育园地
- 临床研究方法学园地
The Bayesian analysis method is a statistical inference framework that updates the probability estimate of a parameter or assumption based on Bayesian theorem combined with prior knowledge and new data. In the Bayesian analysis method, the prior knowledge is expressed as a prior distribution, and the new data are considered through a likelihood function. Combining the prior distribution and the likelihood function can obtain a posterior distribution, which reflects the updated knowledge of the parameters after the new data are observed. Assuming that the prevalence of winter flu is 1%, and the probability of a person getting the flu when there is no other information is 1%, if the incidence of cough among people with known flu is 20 times that of healthy people, then the probability of cough patients getting the flu will grow to about 17%. Here 1% is the prior probability, 20 is the likelihood ratio, and 17% is the posterior probability. If there are data such as fever and blood routine, the posterior probability can be further estimated to improve the accuracy of diagnosis. The advantages of Bayesian analysis methods lie in the ability to quantify the uncertainty of parameters directly and the flexibility to update the estimates of parameters when new data are obtained, providing a flexible and powerful tool to evaluate treatment effects, make decisions, and predict.
本期目次

