DISSECTING THE SPATIAL HETEROGENEITY OF SYNOVIAL INFLAMMATION IN RHEUMATOID ARTHRITIS - PROJECT SUMMARY/ABSTRACT Analyses of rheumatoid arthritis (RA) synovium have shown that the clinical heterogeneity of the disease is in part explained by heterogeneity in which cell types predominate in the synovium. However, we lack an understanding of how these key cell types work together in space, and in particular whether they form “functional units” of co-localizing cells that interact to create specific immune milieus. Identifying whether these functional units exist, how they differ across disease subtypes, and how they relate to the immune aggregates seen in other diseases could improve our mechanistic understanding of synovial inflammation, enable better disease subtypes, and drive new therapeutic hypotheses. Emerging high-dimensional spatial datasets hold the key to addressing this challenge, but analyzing them properly poses significant difficulties. My prior work and preliminary data show that rigorous spatial analysis of inflamed synovium is imminently possible. Specifically, a statistical tool I developed for multi-sample single-cell datasets turns out to be extendable to the spatial realm, where it appears powerful and is able to detect spatial correlates of known RA subtypes even on pilot data. Additionally, I have found that artificial intelligence (AI) can be used to accurately sort small patches of tissue into groups with similar biological content, enabling comprehensive cataloging of the types of cellular aggregates in inflamed synovium and comparison to other diseases. Here, I propose to systematically characterize spatial synovial heterogeneity in RA by applying cutting-edge computational methods to spatial datasets from RA and other autoimmune diseases. In Aim 1, I will use advanced statistical techniques to identify the spatial underpinnings of known RA subtypes defined from single- cell data, and then to define new, spatially informed RA subtypes. In Aim 2, I will use AI to build a map of the different types of immune- and immune-tissue aggregates in inflamed synovium, and then to identify which are also present in other autoimmune diseases. These aims will identify the spatial building blocks of synovial inflammation and relate them to heterogeneity across the spectrum of disease in RA. This proposal will build important new expertise for me in immunology and spatial methods, and it will enhance my existing expertise in clinical rheumatology and scientific leadership. I will pursue it in an excellent training environment with mentors and collaborators who will provide valuable access to key datasets and methodologic expertise. This training will leave me poised to become an R01-funded investigator who uses AI and advanced statistics to improve our understanding of rheumatoid arthritis.