Modeling for Disease Control Improvement
The work of the Smith Lab at the University of Illinois College of Veterinary Medicine
Our work focuses on improving health by finding better ways to intervene in disease processes. We work on all sorts of diseases, from tuberculosis in cattle to breast cancer in women, and all sorts of outcomes, from profit to survival. Across all these systems, one thing remains the same: we want to be able to tell people the best way to meet their goals.
For this research, we rely heavily on two groups of methods: longitudinal data analysis and disease modeling.
Longitudinal Data Analysis: Understanding how diseases change over time is the key to finding what makes them better. We use the tools of longitudinal data analysis (i.e., survival analysis, time series analysis, analysis of repeated measures) to better understand the dynamics of disease.
Modeling: “Model” is a broad term that covers a lot of different areas, and we get into quite a few. Our research uses simulation modeling, risk analysis, and Bayesian network models to explore the potential outcomes of disease systems, and to predict the long-term impact of interventions.
- We are working to find better methods for controlling Mycobacterium avium paratuberculosis (MAP) in dairy herds. MAP causes chronic infections with slowly increasing impacts on dairy production and health. Our data analysis has found evidence of 2 types of infection, progressing and non-progressing. Our modeling work is producing recommendations for the best control programs; it turns out, there is no single ‘best’ control program, and the list of options for a farm are going to depend on the farm’s size and infection level.
- We are working with collaborators to better understand the dynamics of menopause in women as they go through mid-life. Our data analysis has found a number of interesting results, including that quitting smoking can make hot flashes more mild and that moderate alcohol use is related to hot flashes not lasting as long. Our modeling work aims to provide clinicians with a tool for predicting menopausal symptoms and recommending treatments based on a woman’s individual characteristics – personalized medicine meets technology.