Bayesian Methods for Robust Causal Inference and Optimal Experimental Design Under Resource Constraints

My PhD research focuses on developing statistical methods to improve causal inference and optimal experimental design, primarily within toxicology, pharmacology, and public health. Often, experiments in these fields have limited resources or noisy data, making conclusions difficult.
My goal is to develop advanced Bayesian statistical methods that integrate prior knowledge with observed data to improve estimation accuracy and reduce uncertainty in dose-response studies. So far, I have developed Bayesian techniques to better characterize dose-response curves from animal experiments, tested successfully on real-world toxin data (antimony trioxide), and validated through simulation.
I have evaluated optimal experimental designs to efficiently select doses, sample sizes, and group allocations, improving the reliability and effectiveness of research results. Additionally, I'm studying efficient Bayesian techniques to handle complex, noisy biological data. This includes Gaussian process regression models, Bayesian ensemble modeling (combining multiple models for accuracy), and generative machine learning methods (Generative Adversarial Networks) that produce realistic synthetic datasets to strengthen statistical analysis and testing and allow for more better sharing of data between human and animal data for improved estimation.
Moving forward, I will further develop computational tools to speed up Bayesian calculations, extend Bayesian optimization strategies, and apply these techniques to adaptive experimental designs often used in biomedical and regulatory contexts.
My research provides practical statistical and computational tools that allow researchers in health, biomedical testing, and risk assessment to make accurate, reliable decisions, even when facing resource limitations and noisy data.