The treatment of chronic conditions accounted for 58% of the annual healthcare spend in Canada in 2012, primarily through the use of pharmaceuticals. However, these are generally best suited to treat acute diseases, as with chronic use, side effects can accumulate over time while therapeutic effects diminish. Neuromodulation of the Peripheral Nervous System (PNS) represents a promising and adaptable treatment alternative to pharmaceuticals in many cases. Such treatments are still in their infancy and are currently dominated (>99.5%) by devices utilizing open-loop stimulation with clinician-led, manual adjustment. A closed-loop system that responds to peripheral nerve activity and other biomarkers in real time would enable dosesensitive and targeted therapies. However, closed-loop neuromodulation systems face a significant challenge; smart adaptation requires an understanding of how particular nerves encode information to govern the behavior of tissues or organs. New methods must therefore be developed to decode and harness the large volumes of highly complex information transmitted through the PNS. This project will employ the latest findings in machine learning to extract biomarkers from neural data. Semi-supervised training methods will determine how these biomarkers drive physiological responses. The proposed approach will yield methods for robust, realtime calculation of neural biomarkers for targeted nerve stimulation patterns, which will ultimately serve as a data science platform for reliable, chronically implanted neuromodulation devices.