Predictive Modeling of Immune Responses to Pertussis Vaccination. In Data Science & Engineering Master of Advanced Study (DSE MAS) Capstone Projects

Our capstone project focuses on Pertussis, commonly known as Whooping cough, a highly contagious respiratory infection. We explore the challenges and nuances of the two main vaccines: whole-cellular (wP) and acellular (aP). Our research highlights the balance between the safety and effectiveness of vaccines, emphasizing the necessity of ongoing monitoring and research to evaluate how vaccine-induced immunity fluctuates over time in individuals, ensuring sustained effectiveness and safety in public health. The primary goal of the project was to create predictive models for forecasting immune response outcomes following pertussis vaccination, specifically targeting IgG antibody titer levels 14 days post-vaccination, monocyte frequencies one day post-vaccination, and gene expression levels of genes like CCL3 three days post-vaccination. The team utilized a comprehensive dataset of over 500 blood samples from 118 participants, including detailed demographic and immunological profiles. Through rigorous data preprocessing, including handling missing values, detecting outliers, and feature selection, the data was prepared for model building. A variety of models, from simple linear regressors to advanced ensemble learners like Random Forest and Gradient Boosting, were trained and evaluated using cross-validation. The models' performance was assessed using metrics such as R-squared, Mean Absolute Error (MAE), and Mean Squared Error (MSE), with ensemble methods demonstrating superior predictive accuracy. The findings revealed that targeted feature selection and advanced modeling techniques significantly enhanced the predictive power and reliability of the models in understanding and forecasting immune responses to vaccinations.