The Kalman filter model performs at higher specificity and sensitivities for significant disease progression classification due to the greater difference in mean fitted values (i.e. average estimated probability of progression) between progressing and nonprogressing instances.
It is important to note that the average estimated probability of progression of the Kalman filter progressing instances (0.738) is much higher than the average for the raw observations progressing instances (0.498).
The model also considered treatment cessation for each treatment arm, which was extrapolated in the same manner as survival, as well as the probability of progression to AML.
MDS patients have a 20-30% probability of progression to AML and a 40-65% probability of death due to complications and co-morbidities, with the frequency depending on age and comorbidities [1].
A decision analytic model and Markov modeling techniques previously designed [14] were used to simulate the progression of a hypothetical cohort of 1,000 Spanish patients who are initially not on dialysis, representing the relevant stages of the natural history of the disease over time, as well as estimates for probability of progression between the health states related to CKD.
Individuals recovering from tuberculosis, either through treatment or self-recovery, transition back to a state of latency from which they have only a small probability of progression.
Similar to earlier analyses [11], our analysis revealed a lower probability of progression from HIV infection to AIDS within 1 year and within 3 years for female IDUs, compared with male IDUs.
The probability of progression from latent infection to disease is dependent on the duration of infection; here, the risk of progression is greatest within five years of an infection event, and is much reduced after this period has elapsed [7], [8].
For the baseline simulations we present, we assume that the DR strain has a fixed moderate fitness cost which reduces both the probability of transmission and the probability of progression after infection by 20% each; in further analyses, we discuss how an increasing mean relative fitness among DR strains over the course of a drug resistant TB epidemic would affect our results.
The principal determinant of the probability of progression to ESRD is kidney function.
Analysis of the logistic regression fitted values, i.e. the estimated probability of progression, is presented in Table 1.
