1: \begin{abstract}

2: Group testing is a method of identifying 

3: infected patients by performing tests on a pool of specimens collected from patients.

4: For the case in which the test returns a false result with

5: finite probability, Bayesian inference 

6: and a corresponding belief propagation (BP) algorithm are introduced

7: to identify the infected patients from the results of tests performed on the pool.

8: It is shown that the true-positive rate is improved by taking into account 

9: the credible interval of a point estimate of each patient.

10: Further, the prevalence and the error probability in the test 

11: are estimated by combining an expectation-maximization method with the BP algorithm.

12: As another approach, a hierarchical Bayes model is introduced

13: to identify the infected patients and estimate the prevalence.

14: By comparing these methods, a guide for practical usage is formulated.

15: \end{abstract}

16: