Constructing indicators for the prevalence of human immunodeficiency virus and tuberculosis using adjusted maximum likelihood

• Main Authors: Arnan, Abigail, Azajar, Charmine
• Format: text
• Language: English
• Published: Animo Repository 1999
• Online Access: https://animorepository.dlsu.edu.ph/etd_bachelors/16559
• Institution: De La Salle University

The paper involves the deviation of the adjusted maximum likelihood (AML) estimator of the prevalence based from the maximum likelihood (ML) estimator. The AMLE is an improvement of the MLE when it is zero. It estimates the proportion of unobserved number of truly diseased subjects given the number of truly diseased individuals. Using the derived MLE and AMLE, indicators for the prevalence of HIV infection and Tuberculosis were constructed from January to December 1997 in cities and regions of the Philippines, respectively. In HIV infection, the AML indicator showed the improvement of the ML when the prevalence is suspected to be zero. On the other hand, a variation of specificity and sensitivity was done to compare the ML and AML of Tuberculosis. When sensitivity = 0.8 and specificity = 0.714, the ML indicator is zero and the AML indicator is negative while its limit is zero. On the other hand, if sensitivity = 1.0 and specificity = 0.98, the ML and AML have positive values and are equal to one another.