Immigrant enclaves are central to the U.S. immigration canon and are hypothesized to play an instrumental role in shaping the health and well-being of immigrants and their descendants. Where immigrants live is understood to be a key barometer of their prospects for success and is closely tied to the challenges they and their offspring will confront as they begin their lives in the U.S. But the rigor of existing quantitative assessments of the impact of enclave residence is limited by the available methods used to operationalize immigrant enclaves. We propose to apply recent developments in machine learning (ML) to improve our ability to empirically characterize the variety of neighborhood types in which immigrants cluster in order to ultimately assess how they impact immigrant health. The need for new measures is underscored by massive changes in immigrant settlement patterns at the turn of the 21st century that have challenged the analytic utility of past conceptualizations of enclave communities. Geographic diversification of immigrants, the rise of ‘ethnoburbs,’ and the suburbanization of poverty are all trends that demand new theorizing and new measures of immigrant residential context. Our aim is to leverage advances in ML to produce a more thorough accounting of the variability that exists across immigrant neighborhoods, precisely the type of variability that may be consequential for health. Data will come from the American Community Survey (ACS) and U.S. censuses.