Proactively Predicting Blood Glucose from Intracranial Electrophysiology
Here, we undertook a unique opportunity to enroll patients undergoing clinical brain monitoring into a study integrating simultaneous glucose measurements. This allowed us to sample multiple putative brain regions with high spatiotemporal resolution and correlate neural activity to glucose changes over a semi-chronic duration. Coverage of the hypothalamus bilaterally as well as other subcortical and corticolimbic regions in one exceedingly rare patient allowed us to study one of the most implicated regions in central glucose homeostasis. Through this dataset, we were able to extensively characterize the relationship between central nervous system signals and peripheral glucose changes on multiple timescales and build a neural decoder capable of predicting glucose levels both in the present and hours into the future. Our work provides a key advance in the understanding of how human intracranial activity couples to peripheral metabolic activity.