Abstract:
Mycetoma is a chronic granulomatous neglected tropical disease. The danger of mycetoma is that it painless at first, but takes the patient to late stages. Moreover, it is highly recurrent. A recurring events model has become one of the most important tools for analyzing recurring events data. A major goal of this study is to model recurrent mycetoma events and to compare parametric (Weibull, exponential, and log logistic) models among them and parametric and semi parametric models between them. It will also compare the recurring events models using Akaike and Schwartz-Bayesian criteria in order to find the most appropriate model for analyzing recurring events. In this study, all 171 cases of recurrence of mycetoma were included in the data gathered at the Mycetoma Research Center in Khartoum, Sudan, between 1991 and 2021.
(Analyzing data was carried out using frequency tables, custom tables, quartile tables, life tables, Kaplan-Meier models, and parametric and semi parametric models). As a result of the comparison of the models, log logistic model was the best the model based on Akaike and Schwartz-Bayesian criteria, whereas Kaplan Meier results showed that the most significant factors are (age, duration of the disease, veins, periosteal reaction, organism, ultrasound, treatment and length of treatment). At the end Four models were built in different ways for modeling recurrent events, the first model run by divided survival time into (start and stop time), second model run using (stratum) , third model run by adding cluster variable (id) , a parametric weibull model is run with a gamma distributed shared frailty component for the fourth model. The results of comparison showed the second and third models the best based on (significant value) tests, while the fourth model the best using Akaike and Schwartz-Bayesian testing. This final model of mycetoma is recommended for predicting recurrences, as well as conducting additional studies with Cox's stratified methods and other parametric methods.
