Conway-Maxwell-Poisson model fitting to HFMD data in Malaysia

Since the first Hand, Foot and Mouth Disease (HFMD) outbreak occurred in Sarawak, Malaysia in 1997, the number of reported cases has remained in a cyclical pattern. Numerous HFMD research involve clinical and laboratory findings. However, there is limited research that models the distribution of the HFMD dataset in Malaysia using a count data model fitting. This study aims to demonstrate the flexibility of the Conway-Maxwell-Poisson (COM-Poisson) regression model. All daily reported cases of HFMD from 2009 to 2019 were analysed and presented by each state in Malaysia. A normal Poisson was extensively used, but it has limitations in terms of the equi-dispersion assumption. Thus, the performance of the COM-Poisson model was investigated and compared to the Poisson model by looking at the goodness of fit (log-likelihood, AIC, and BIC) test values. The results showed that COM-Poisson models fit the HFMD dataset well with lower AIC and BIC values compared to the Poisson model. The parameter coefficients' estimated values also indicated smaller values than in the Poisson model. Given the versatility of the COM-Poisson distribution, it is effective for statistical applications and procedures, such as modelling count data. In addition, the potential for related research was also being examined to improve the accuracy of the predictive model.