Bayesian modeling for multivariate randomized incomplete block design:application in sperm biology researches

• Published in: Journal of biostatistics and epidemiology Vol. 1; no. 1/2
• Main Authors: Abbas Rahimi-Feroushani, Mohammad Chehrazi, Keramat Nourijelyani, Reza Omani-Samani
• Format: Journal Article
• Language: English
• Published: Tehran University of Medical Sciences 01.10.2015
• Subjects: Bayesian estimation; cryopreservation; Markov chain Monte Carlo methods; sperm biology
• ISSN: 2383-4196; 2383-420X

Background & Aim: The aim of the current study was to investigate the advantages of Bayesian method in comparison to traditional methods to detect best antioxidant in Freezing of human male gametes. Methods & Materials: Semen samples were obtained from 40 men whose sperm had normal criteria. A part of each sample was separated without antioxidant as fresh and the remaining was freezed with and without antioxidant. Taurine (in concentrations of 25 mm and50mm) and cysteine (5mm and10mm) as antioxidants were prepared as intervention. Traditional results were obtained from randomized incomplete block design and compared with Bayesian results in their ability to find the significant difference among our groups. Using Markov chain Monte Carlo algorithm within the WinBUGS software, we developed a Bayesian approach to estimate the protective effect of antioxidant against inverse effect of freezing on the quality of sperm. Results: Classic method could detect the significant difference just in cycteine10mm for viability which was confirmed by Bayesian method. In Bayesian method, in addition to results from classic method, we could find the significant improvement in abnormality: cysteine 10mm, protamin deficiency: taurine 25 mm and10 mm, viability: cysteine 10mm, DNA fragmentation: cysteine 10mm which all of them was interested in clinically, but could not be proved by the traditional methods. Conclusion: Bayesian approach in sperm biology research can be considered as a good replacement of the traditional methods for estimation. Using this method, we can solve complex and intractable statistical models. Future researches should be done to confirm our suggestion.