Abstract Long-term cognitive impairment affects more than 70% of sepsis survivors, but the underlying mechanisms remain unknown. Though widely hypothesized, evidence of blood-brain barrier (BBB) dysfunction in septic patients is limited by practical barriers to diagnostic studies in critically ill subjects. While BBB breakdown and cognitive impairment are seen in animal models of sepsis, the complexity of sepsis in vivo and differences between animal and human responses means that animal models cannot unambiguously identify the circulating factors that cause brain injury in human sepsis. Therefore, we propose to develop the µSiM-hNVU as an `on-chip' platform featuring a human iPSC-derived neurovascular unit (NVU; brain microvascular endothelial cells, pericytes and astrocytes). The `blood side' will allow the flow-based introduction of blood- borne cells and molecules with known or hypothesized roles in sepsis related brain injury, and the `brain side' will feature iPSC-derived microglial cells serving as a reporter of the brain inflammatory status. The human NVU will be built on a device platform ? the µSiM ? featuring ultrathin silicon nanomembranes that provide for unhindered solute exchange between `blood' and `brain' compartments and glass-like optical quality for live cell imaging and high-resolution microscopy. In the R61 phase, the device platform will be advanced for ease-of- use including `plug-and-play' modules for flow and barrier measurements (TEER, diffusion), and compatibility with a small-volume, digital-ELISA assay for secreted proteins. The µSiM-hNVU will be validated with functional assays of blood-brain barrier (BBB) function, protein expression studies, and transcriptional analysis. We will also build a iPSC NVU in which each cellular component of the NVU carries the ApoE4 allele. The expression of the ApoE4 lipoprotein drives BBB dysfunction by a known pathway and increases the risk of cognitive impairment in humans and animals experiencing brain inflammation. We will use the ApoE4-NVU as a `diseased BBB on a chip? which we hypothesize will show enhanced vulnerabilities to candidate mechanisms of brain injury identified by our team and others. Specifically, we will test the hypotheses that 1) pre-activated monocytes invade the brain and drive microglial activation; 2) the damage associated molecular pattern (DAMP) complex S100A8/A9 drive BBB breakdown to promote leukocyte infiltration and neuroinflammation; and 3) circulating factors that degrade endothelial glycocaylx (e.g., heparinase) or contribute to systemic inflammation (cell-free hemoglobin) promote CNS infiltration of leukocytes and subsequent neuroinflammation.