FionaHuini/tb-kenya-transmission-model

Compartmental TB transmission model calibrated to Kenya WHO data (2000–2022) # TB Transmission in Kenya: What Will It Take to Meet the 2030 Target? Kenya has one of the highest TB burdens in Africa. Despite decades of investment in its national TB programme, incidence remains far above the WHO End TB 2030 target. This project asks a simple but consequential question: under realistic intervention scenarios, how close can Kenya actually get? To answer it, I built and calibrated a compartmental transmission model using WHO Kenya incidence data and Kenya-specific parameters from the published literature. The model is deliberately simple with five compartments, fixed parameters, single-point calibration. I think that honesty about what a model can and cannot do is more useful than complexity that obscures its limitations. ## The Questions This Project Addresses 1. Is Kenya's TB epidemic self-sustaining without intervention, and what does the basic reproduction number tell us about that? 2. What does Kenya's observed post-2000 incidence decline tell us about what actually drove it, biology or programme improvement? 3. Which programmatic parameter: treatment initiation, transmission reduction, or both combined, offers the greatest leverage for reducing TB burden by 2030? 4. Which biological and programmatic factors most strongly determine TB incidence in Kenya, and what does that mean for where the health system should focus? ## What I Found Kenya's calibrated R0 is 1.10. TB persists without intervention. The model cannot reproduce the observed post-2000 decline in incidence, and I think that is informative rather than a failure. It suggests the decline was driven by improving treatment coverage over time, not by changes in transmission biology. A fixed-parameter model is not designed to capture that. None of the three intervention scenarios reaches the End TB 2030 target of 76 per 100,000. Even combined treatment scale-up and transmission reduction projects 216 per 100,000 by 2030, nearly three times the target. The sensitivity analysis …