Models for Epidemic Control: Targeting Tuberculosis Patients for Treatment

Abstract:  

Tuberculosis (TB) is the ninth leading cause of death worldwide, killing over 1.3 million people in 2016.  Recent efforts to combat the disease have included public outreach campaigns to educate potential patients on TB symptoms and encourage them to seek treatment.  We inform this effort by integrating multiple data sources with a theoretical framework of disease transmission, then developing a novel algorithmic procedure to optimize which age groups should be targeted in a TB outreach campaign for India. We find that the optimal allocation of public health funds varies by the duration of the time horizon, where outreach efforts should be concentrated in specific age groups if a shorter time horizon is used and more diffuse targeting across multiple age groups should be used for longer time horizons. 

Once in the healthcare system, patients with drug resistant disease may need different treatment than those with drug-sensitive disease.  However, identifying these patients may be difficult since tests to determine disease strain may be time-consuming or costly.  In a second project, we develop a model using POMDP and simulation techniques to identify when and which first-line tuberculosis patients are most likely at risk for drug resistance and should be screened to reduce costs and increase health outcomes.