|
Abstract:
|
Leprosy is a neglected tropical infectious disease that presents differently in patients depending on the ‘state’, described along a spectrum from tuberculoid to lepromatous. Uniquely, patients can transition over time between these disease states, with important implications for treatment and control. There are five recognized disease states between the two polar forms. Acute inflammatory reactions can occur before, during, and after treatment, and can lead to lasting disability. Previous studies have examined leprosy disease transmission between individuals, but we seek to investigate within-individual disease progression. We propose a novel hierarchical Bayesian model analyzing disease progression over time. For a given patient/time, the transition probabilities follow a multinomial likelihood. The model is temporally structured by conditioning on an individual’s previous disease status and incorporates fixed and time-varying covariate data. Residual individual-level variation can be accounted for through random effects. Importantly, this longitudinal approach allows us to model the probability of reactions over time and can be an important prognostic tool for clinicians
|
