A Neonatal Intravenous Monitor Prototype

Abstract

An intravenous (IV) infusion monitoring system prototype H2neO has been developed for low drip rates (25–400 drops/hr) using a macrodrip infusion set (gtt/drop factor of 20 drops/ml), corresponding to low IV infusion rates (1–20 mL/hr). We demonstrate H2neO as an accurate, low-cost, gravity-based solution that can be used in low-resourced settings for neonates. H2neO uses an infrared emitter, detector, interface circuit, and signal processor implemented on an Arduino-based microcontroller to detect drops in a standard macrodrip infusion system. The infrared signal was preprocessed using a median filter to establish the baseline (no drop) condition and an 8.5% threshold to determine the initial presence of a drop in the IV drip chamber. Using signal preprocessing alone, the monitoring system underestimated the drip rate and flow rate by a maximum error of 1.33% and overestimated it with a maximum error of 0.91%. Using a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">k</i> -means approach to classify drops into “normal” and “noisy” clusters and then considering drip rates estimated from only the “normal” cluster, underestimation errors improved at all infusion rates tested while overestimation errors improved at half of the infusion rates tested. The average detection error over 1–20 mL/h infusion rates was +/- 0.5% compared to +/- 1% for commercially available monitors and +1.8% for recent results reported in the literature. This approach, based on both signal preprocessing and analyzing the shape of each drop, is promising for implementation on a standalone, microcontroller-based drip-based IV infusion monitoring system for neonates and other populations, including adults.