UC Riverside

The immune system plays a vital role in protecting our bodies from a wide range of pathogens. As we continue to identify new diseases that challenge our immune system, im- munotherapy has gained prominence as a potential medical solution. It utilizes or enhances the body’s own immune system to detect and destroy cancer cells. Immunotherapy presents advantages over traditional treatments, although its successful application demands a more extensive understanding of the immune system. While progress has been made, our un- derstanding of the immune system, specifically the innate immune component, is still not comprehensive.Some key questions that pique my curiosity include: How do innate immune cells distinguish between self and non-self targets? When transplanted to a different host, how do these cells adapt to their new surroundings? Can this process be elucidated through equations or quantitative models? I have always been deeply interested in exploring these questions. In this thesis, we apply statistical methods to tackle these issues. We develop a quantitative model to elucidate the regulatory mechanisms governing the activity of one type of innate immune cell, namely natural killer cells. We develop a model to simulate the interactions between natural killer (NK) cells and their target cells, illuminating how NK cells learn to identify unhealthy signals from their environment. We apply our model across a range of experimental scenarios, showcasing the algorithm that we have designed to mirror specific experimental settings and delving into the relationship between model and experimental parameters. Further, we provide evidence for the efficacy of our model by demonstrating that it is possible to identify rational values for model parameters that yield accurate estimations of the experimental data. This effectively validates our approach. Progressing from this foundation, we present a high-dimensional extension of our model, contributing insights into immune protection at a population level. A key part of our discussion involves the distribution of receptor numbers on natural killer cells and the benefits that this distribution bestows. Finally, we put our model to the test in the arena of anomaly detection. Through this exploration, we display the versatility and applicability of our mathematical framework, proving its potential to address other complex, real-world problems.