Multi-Modal Local Physiological Sensing at the Intravenous Catheter Insertion Site : Towards Automated IV Infiltration Detection

Intravenous (IV) catheter therapy is prevalent in hospital and clinical settings, delivering necessary fluid, nutrients, or pharmacological treatments to patients. An IV is typically inserted by a trained vascular access team member in a peripheral vein such that the delivered fluid enters the vasculature directly and is distributed through normal cardiovascular mechanisms to desired regions within the body. The inadvertent leakage of fluid outside the vascular space into the surrounding tissue can occur due to movement of the catheter tip out of the vein with patient motion, or from mechanical or chemical damage to the venous wall. This complication is referred to as an IV infiltration or extravasation and is considered a medical emergency as it can result in tissue damage or even necrosis for the patient. Standard of care for detecting an infiltration involves subjective evaluation by nurses or caregivers: specifically, a "look, feel and compare" approach is applied to detect swelling or changes in skin temperature associate with the fluid in the extravascular space. Our team has engineered a wearable, multi-modal sensing system for detecting local physiological changes around the IV catheter insertion site; we monitor electrical bioimpedance spectroscopy (BIS) for quantifying extravascular fluid, skin temperature at two positions around the site, and motion with two inertial measurement units. We have successfully deployed the system at Children's Healthcare of Atlanta and performed initial feasibility studies in a total of 9 pediatric participants, and have developed algorithms for reducing variability in the BIS measurements associated with motion. While no subject has had an IV infiltration yet, we demonstrated high quality measurements of all parameters in the recordings, and that motion correction of the BIS signals reduced variability when the IV catheter was known to be successfully delivering fluid to the veins. This study sets the foundation for future work where we aim to demonstrate automatic and accurate detection of IV infiltration in pediatric patients with our wearable system and associated algorithms.