Fig. 2
From: Adaptive bandit algorithms increase efficiency of mobile tuberculosis screening programs
Five-state Markov chain model of tuberculosis states, applied at the neighborhood level. The states represent (1) TB positive (TB+) and eligible for mobile screening (TB + unobserved), (2) TB negative (TB-) and eligible for mobile screening (TB- unobserved), (3) screened at a mobile unit and diagnosed with TB (TB + observed), (4) screening at a mobile unit and not diagnosed with TB (TB- observed), and (5) those no longer considered eligible for screening at the mobile units. Individuals in the community develop TB disease at a rate of \(\gamma\); when the screening van is available (at = 1), TB + individuals are screened at a rate of s+, and TB negative individuals at a rate of s-. d represents the rate at which TB + individuals transition directly into State 5 and are no longer expected to present at a screening location. See Table 1 for a descriptive list of these model parameters along with their estimated values.
