Convolutional Neural Networks (CNNs) have traditionally been used for image analysis and computer vision and are known for their ability to detect complex patterns in data. This report studies an application of CNNs within bioelectronic medicine, namely predicting blood glucose levels using nerve signals. Nerve signals and blood glucose levels were measured on a mouse before and after administration of glucose injections. The nerve signals were measured by placing 16 voltage-measuring electrodes on the vagus nerve of the mouse. The obtained nerve signal data was segmented into time intervals of 5 ms and aligned with the corresponding glucose measurements. Two LeNet-5 based CNN architectures, one 1-dimensional and one 2-dimensional, were implemented and trained on the data. Evaluation of the models’ performance was based on the mean squared error, the mean absolute error, and the R2-score of a simple moving average over the dataset. Both models had promising performance with an R2-score of above 0.92, suggesting a strong correlation between nerve signals and blood glucose levels. The difference in performance between the 1-dimensional and 2-dimensional model was insignificant. These results highlight the potential of using CNNs in bioelectronic medicine for prediction of physiological parameters from nerve signal data.

Obesity is pandemic. Pharmacological treatment development depends on modeling the regulation of feeding, particularly by free fatty acids (FFA). Most models have been employed in the rat in vivo, and show FFA-stimulated intestinal satiety signals are dependent on the fat’s acyl chain-length, involve cholecystokinin (CCK) secretion, and are mediated by vagal afferents. I hypothesized that an in vitro mouse model could be employed, with sensitivity to measure afferent responses to nutrient stimuli. Male C57BL/6N mice were killed, the intestine harvested en bloc, and a jejunal section dissected with neurovascular mesenteric arcade emanating centrally. The tissue was placed in a Krebs-superfused chamber, the lumen cannulated with the outlet open to drain, and Krebs or other mediators were continuously perfused intraluminally. The dissected afferent nerve was placed in a suction electrode for extracellular recording. Afferent responses to distension and the perfusion of mediators (e.g. CCK or FFA) were tested. Preparations from normal mice (no surgery), or from mice following chronic subdiaphragmatic vagotomy or sham operation, were used to assess vagal afferent contributions. Luminally-perfused CCK (100 nM) increased afferent firing. This response was abolished with the CCK-1 receptor antagonist lorglumide (10 µM). The short-chain fatty acid (SCFA) sodium butyrate (30 mM) potentiated firing. The long-chain fatty acid (LCFA) sodium oleate (1-300 mM) activated concentration-dependent firing (EC50=25.35 mM) that was significantly greater at 30 mM than that evoked by butyrate. Lorglumide (30 µM) abolished the oleate (30 mM) response. The L-type Ca2+ channel (LTCC) inhibitor nicardipine (3 µM), intraluminally, potentiated the oleate response, while bath application abolished it. Vagotomy attenuated the oleate response. Vagotomy abolished the intraluminal CCK (100 nM) response, and attenuated the response to bath-superfused CCK. These findings support FFA chain-length-dependent mesenteric afferent activation and CCK involvement in oleate-induced firing, and suggest LTCC mediation of excitatory and inhibitory oleate response transduction pathways. The murine oleate response was shown to be mostly vagally-mediated, with some spinal contribution, and both vagal and spinal contributions to CCK responses were suggested. These data provide a basis for further investigation in vitro of cellular and molecular mechanisms of afferent satiety signals, and ultimately of obesity pathogenesis. / Thesis (Master, Physiology) -- Queen's University, 2010-06-29 15:56:08.387