SELECTIVE MODULATION OF AUTONOMIC FUNCTIONS VIA CLOSED-LOOP VAGUS NERVE STIMULATION<S>​</S>

The vagus nerve is a major component of parasympathetic regulation and its electrical stimulation has demonstrated significant therapeutic potential across a range of autonomic conditions, including cardiovascular, inflammatory and metabolic disorders. However, its clinical translation remains challenged by unselective stimulation often leading to off-target effects. In light of this, this project investigates the potential of cervical vagus nerve stimulation (VNS) to selectively engage vagal pathways involved in autonomic control.

To address this challenge, we performed extensive in vivo experiments in five acute porcine models. We focused on three vagus-mediated physiological signals: heart rate, breathing rate, and laryngeal activity. We first aimed to improve spatial selectivity by using an in-house-developed neural interface combining extra- and intraneural active sites. We then developed a Bayesian optimization framework based on Gaussian processes to identify stimulation paradigms for selective neuromodulation. The optimization framework was initialized using a two-step strategy after performing mapping of stimulation sites to physiological responses for each individal pig. First, we identified potential subject-specific functional organization within the nerve. Second, we calibrated algorithm parameters doing an offline population-level analyis of mapping data from all pigs.​
​
After validating correct interface implantation, channel-specific physiological responses were observed, highlighting functional maps. With data-driven algorithm initilization, Gaussian process-based Bayesian optimization converged toward more selective stimulation paradigms, as validated in both offline and online procedures. These findings represent an first step toward elucidating VNS paradigms that selectively target autonomic pathways and further highlight the potential of VNS as a therapeutic approach for several pathological conditions.