Research
Research Summary: Leveraging Large-Scale Community-Based Data to Promote Equitable and Inclusive Access to Healthcare through Optimal Intervention and Policy Decisions
Communities Impacted to Date: Underserved Communities in Subsaharan Africa and United States (Georgia, Michigan, New York, Massachusetts, overall US)
Addressed Health Issues to Date: Pediatric Asthma, Mother-to-child transmitted HIV, Invasive Pneumococcal Disease, Opioid Use Disorders, Pregnancy and Postpartum Care, Mental Health
Methods: Stochastic Processes (Discrete-time Markov chain, Continuous-time Markov processes), Simulation (Discrete event, Microsimulation, Agent-based, System dynamics, Differential equations), Optimization for Resource Allocation (Dynamic programming, Stochastic optimization), Predictive Analytics (Machine learning, Deep learning)
Example of Research Projects:
Asthma
Asthma is one of the most common chronic pediatric conditions in the US. There is no cure for asthma; however, the severity of the disease may alter over time. Avoidance from triggers and adherence to medications can help to control disease and prevent exacerbation. Even though guidelines have recommended asthma interventions, their net effects are unknown. By combining predictive modeling and data mining methods, we develop models to predict the impact and return-on-investment of pediatric asthma interventions.
Invasive Pneumococcal Disease (IPD)
Streptococcus pneumoniae is a bacterial pathogen causing various clinical conditions, including meningitis, septicemia, and pneumonia. Although high vaccination coverage rates, cases of vaccine-type IPD after completing the recommended course of immunization (vaccine failure) have been reported. In this study, we predict the vaccine failure probabilities for various vaccine schedules and ages.
Depression
Depression is a mental health condition affecting more than 16 million adults in the United States, with approximately 64% of those experiencing significant impairment. The implications for society are substantial, with low productivity, comorbidities and their associated cost, and even higher mortality. In this research, we built a natural history model for depression and evaluated the routine screening schedules for the US adult population.