This research project aims to identify various distinct disease patterns or sequences that occur throughout people’s lives. While there has been some prior research in identifying patterns of disease, systematic approaches to identifying commonly occuring patterns of disease diagnoses have not be established, and linking genetic information to these patterns have not been explored. Over the course of four years, we plan to develop novel longitudinal clustering approaches to identify commonly occuring patterns of health outcomes. In addition, we will look for genetic variation that is enriched in individuals exhibiting certain patterns, and this may indicate of distinct genetic subtypes of disease. Furthermore, we also plan to use use machine learning and deep learning to predict the next likely diagnosis based on disease trajectories. This will include methods such as polygenic risk scores, and we also plan to develop novel deep learning architectures that can accomodate genetic data to predict disease classifications or disease trajectories.
