Wan Muhamad

A Way of Selecting the Right Statistical Model for Handling Overdispersion and Excess Zeroes

This study aims to find the risk factors of Diabetes Mellitus (DM) and to find the best model among Poisson, Negative Binomial (NB) and Zero-Inflated Negative Binomial (ZINB) regression models. In the count data, the existence of overdispersed data is a common situation for modeling approach. Overdispersion occurs when the data has greater value of variance compared to its mean. The Poisson regression model is a good starting step to model the data but does not account for overdispersion. Hence, NB regression model provide a better way to handle this case. Although it works well, but its inclusion of dispersion parameter seems to increase the probability of zero counts. As a result, we suggest applying ZINB regression model as it capable to handle both overdispersion and excess zeroes. Under the test of Akaike Information Criterion (AIC), Likelihood Ratio (LR), Vuong and Clarke, the ZINB regression model was chosen as the best model. From this analysis, we found that 11 of the parameters were the risk factors that significantly associated with DM at p 0.05. Meanwhile, the respiratory system is the major contributor to the problem of DM.

Exposure and Factors Influencing the Progression of Diabetic Nephropathy AmongstType II Diabetes Mellitus Patients in Seremban, Negeri Sembilan

The case of diabetic nephropathy often arisesamong people with type 1 or type 2 diabetes. The incident of diabetic nephropathy has increased dramatically in the last decade. Increasing case of diabetic nephropathy had worried to most of diabetic patient. A cross-sectional study was conducted andfocused on two areas:modeling the applied logistic regression and todetermine the factors that influence diabetic nephropathy mellitus using structural equation modelling in Seremban, Negeri Sembilan.Three hundred eighty three diabetic patients were selected from Seremban. Of those who had diabetes, 172 (44.9%) patients were female and 211 (55.1%) were male.Finding from the previous research, the earliest age tostart diabetes mellitus is at 20 years old. Factors found to be significantly associated with diabetes mellitus on multivariate analysis werebeta blockers (OR= 3.942, 95% CI= 0.422, 36.854), serum creatinine (OR= 1.057, 95% CI=1.023, 1.093), age (OR= 0.917, 95% CI=0.821, 0.97), blood glucose (OR= 1.564, 95% CI= 1.091, 2.241), gender (OR= 20.396, 95% CI= 0.963, 431.892). In this paper, we found that age, being male,taking Beta-blockers having blood glucose level and serum creatininereported as influencing factors of type II diabetes mellitus.

Influence By Injection Drug Use In Patients With Hiv-Tb

Every year reported a patient with HIV is increasing. HIV can infect humans, through sexual whether heterosexual or homosexual, infection through contaminated blood flow, such as sharing drug injection equipment and infection from mother to child. HIV / AIDS, TB and injection drug use (IDU) contribute to the spread of the epidemic and increased mortality rates in communities around the world. TB is the leading cause of death among HIV-infected IDUs. In this present paper, we present the results that gained from binary logistic regression method and used to model the relationship between the ordinal outcome variable. The significant variables are chosen based on the p-value criteria which associated with the significant level of model that lies at the level of  = 0.25. Logit determination and the correlation between the variables are also discussed for further analysis. The use of mathematical software PASWand AMOS version 18 is applied in this research as an alternative calculation procedure derived from the methodology. Among the factors involved in this study were age, gender, race, level of CD4 content of less than 200/mm3, heterosexual, receiving blood infected with HIV, smoking, hepatitis, pneumonia and toxoplasmosis. This study found that there were six significant factors were age, gender, levels of CD4 less than 200/mm3, heterosexual, smoking and pneumonia. For structural equation modeling showed that the critical ratio for the diagnosis of HIV and TB patient through HIV_TB: IDU and CD4 was significant level at 0.05 and have a strong relationship. Overall, the results showed that age, gender, level of CD4 content of less than 200/mm 3 , heterosexual, smoking and pneumonia were contributing factors to TB infection in people with HIV-positive through injection drug use (IDU) in Kelantan. The government and relevant parties should provide ongoing effort to increase awareness about the dangers of social relations and illness with TB and HIV-positive IDUs by conducting campaigns, talks and tighten laws involving social problems.