This thesis presents a process to create 3D thermal models from multi view images captured with low cost devices, a smartphone and low cost thermal camera. 3D thermal models contribute to a data-driven assessment of chronic wounds and improve the limitations of manual measurements and subjective visual observation. Leveraging color and thermal imaging, realistic 3D thermal models are created from a limited set of images, demonstrating their effectiveness in wound assessment in a clinical study, emphasizing portability and low-cost applicability.

In this work, I explore Bayesian approaches for estimating disease prevalence when accurate diagnostic tests are unavailable. Simulation studies assess the performance of two methods, enabling estimation of chronic kidney disease prevalence in Peru using the CRONICAS cohort study.