Start Submission Become a Reviewer

Reading: Bayesian linear regression model for method comparison studies

Download

A- A+
Alt. Display

Research Articles

Bayesian linear regression model for method comparison studies

Authors:

S.M.M. Lakmali ,

University of Peradeniya, LK
About S.M.M.
Postgraduate Institute of Science
X close

L.S. Nawarathna,

University of Peradeniya, LK
About L.S.
Department of Statistics and Computer Science
X close

P. Wijekoon

University of Peradeniya, LK
About P.
Department of Statistics and Computer Science
X close

Abstract

Method comparison is an essential area related to clinical science and study to compare a new method with an existing method to check if a new one can replace with an existing method. This study proposes an efficient methodology for homoscedastic measurements to evaluate the agreement between two methods using Bayesian inference. This proposed model introduces an accurate, model-fitting easiness, less time required, and an assumption-less model. Simulation is used to assess and compare the finite sample performance. Simulations were carried out, and the coverage probabilities and credible intervals were calculated for each trial. Coverage probabilities of model parameters, alpha, and beta, imply that those are between the credible interval with 96% and 96.5%, respectively. It is observed that the coverage probabilities are decreasing with the increase of sample size. The proposed methodology is then used to analyze the Cardiac Ejection Fraction data. The best-fitted model was selected using the minimum error values and evaluated the agreement between the two methods using that model. The proposed model was chosen as the best model, considering the two methods have good agreement. The proposed model performed well, specifically with the small sample sizes.
How to Cite: Lakmali, S. M. M., Nawarathna, L. S., & Wijekoon, P. (2022). Bayesian linear regression model for method comparison studies. Ceylon Journal of Science, 51(1), 37–42. DOI: http://doi.org/10.4038/cjs.v51i1.7977
Published on 14 Mar 2022.
Peer Reviewed

Downloads

  • PDF (EN)

    comments powered by Disqus