ID: 212135 Noninvasive Human Immunotyping Through Machine Learning Using the Autonomic Nervous System

The extreme diversity of the human immune system, forged and maintained throughout evolutionary history, has to date provided a potent defense against pathogens [1]. Intrinsically variable autonomic neuro-immune regulation via β-adrenergic receptor and α7-nicotinic acetylcholine receptor activity is controlled by polymorphic genes and environmental sensors [2, 3]. While most healthy humans have the capacity to activate various types of immune cells, interferons, inflammasome activity, and vagal anti-inflammatory reflexes, individuals differ in the degree to which they are primed for each functional configuration. Virulent emerging pathogens inherently circumvent these functional immune configurations and can impose a dysregulated hyperinflammatory response [4]. Here we detect host-dependent baseline autonomic gradient drive (AGD) bias through high temporal resolution physiological monitoring. Similar to immune bias, AGD bias shown to predict current inflammatory status as well as severity of future inflammatory sequela and disease.