Bayesian Meta-Analysis

Meta-analysis which is widely used in evidence-based medicine, is a statistical method for combining multiple studies with the same research purpose and quanti-fying the results of comprehensive evaluation. When the probability of some event is small, events may occur in either or both arms of the study, which is called the sparse data. For processing the data sparseness, The most frequently used method is the continuity correction or directly the Bayesian model.Based on the Bayesian method, this paper gives the link distribution:when the prior distribution is the Jeffreys prior, i.e.Beta(1/2, 1/2). In general, We given the link distribution:when the prior distribution is Beta (?, ?). Finally we give the experimental results,and discuss the sensitivity of the results with t varying within a selected range of values.