The vagus nerve carries information about physiological processes including inflammation, metabolism, and cardiovascular function, making it an attractive target for next-generation bioelectronic medicine. However, analytical tools for peripheral nerve recordings remain underdeveloped compared with those used in neuroscience, and conventional approaches such as spike sorting may be poorly suited to the sparse and asynchronous activity observed in vagus nerve recordings. This project will develop and evaluate machine learning methods for identifying physiological states from multichannel vagus nerve recordings obtained using high-density electrode arrays. Rather than relying on predefined labels or extensive spike sorting, the framework will use data-driven analysis of temporally aggregated neural activity to detect and compare physiological states across animals and experimental conditions. High-performance computing resources are required to process large electrophysiological datasets, train and validate machine learning models, and systematically compare analytical approaches for decoding vagal neural activity.