Abstract
Over the past few decades, digital twins have grown in popularity as a tool in simulation and data visualization. Digital twins have shown themselves to be a powerful tool in the manufacturing and construction industry and have more recently gained intrigue in the healthcare industry, especially in collaboration with artificial intelligence. With the introduction of digital twins into healthcare being so fresh, there are a lot of use cases for digital twins that have yet to be implemented in any way. Digital twins allow for interactive, detailed, and easy-to-understand visualizations of a patient’s well-being in the form of a virtual 3D model of a patient or a related body part to help bridge the gap between professionals to patient understanding. This thesis discusses the current state of digital twins in healthcare data visualization, focuses on the development of a method to generate digital twins from longitudinal randomized control trial (RCT), and expands upon the uses of this algorithm by applying it to predicted versions of participants derived from the NIH data using linear regression, lasso linear regression, and random forest regression. Additionally, it uses the harmonized longitudinal data to create scenes with these generated digital twins that help emphasize each digital twin’s unique characteristics. The digital twin scenes provided a comprehensive display of a digital twin generated from harmonized longitudinal data. This method serves as a basis for the automated development of digital twins using any dataset that contains biomarker data on its participants. The potential future applications of this technology allow for easy-to-understand displays of a patient’s health status at any point in their healthcare journey.