AiRDose: Developing and Validating an Augmented Reality Smartphone Application for Weight Estimation and Dosing in Children

Objectives Inaccurate weight estimation is a contributing factor to medical error in pediatric emergencies, especially in the prehospital setting. Current American Heart Association guidelines recommend the use of length-based weight estimation tools such as the Broselow tape. We developed the AiRDose smartphone application that uses augmented reality to provide length-based weight estimates, as well as medication dosing, defibrillation energy, and equipment sizing recommendations; AiRDose was programmed to use Broselow conversions to obtain these estimates. The primary objective was to compare the length estimated by AiRDose with the actual length obtained by the standard tape measure. The secondary objectives were to compare the estimated weights and critical medication doses from AiRDose with current established methods. Methods In this prospective validation study, lengths and estimated weights were obtained for children presenting to 2 emergency departments using AiRDose, Broselow, and a standard tape measure; actual weight was recorded from the patient chart. Using the AiRDose estimated weights, hypothetical doses of epinephrine and lorazepam were calculated and compared with doses recommended via Broselow and to actual weight-based doses. Spearman rank correlation coefficients were calculated. We defined an acceptable difference of 20% between AiRDose and standard measurements as clinically relevant. Results Five hundred forty-nine children (mean age, 4.8 years; standard deviation [SD], 2.9 years) were recruited. There were 99.6% of AiRDose lengths within a 20% difference of tape-measure lengths. There was a significant correlation between AiRDose and tape-measure length measurements (r = 0.989, P < 0.0001), and between AiRDose and Broselow weights (r = 0.983, P < 0.0001) and AiRDose and actual weights (r = 0.886, P < 0.0001). AiRDose lorazepam and epinephrine doses correlated significantly with Broselow lorazepam (r = 0.963, P < 0.0001) and epinephrine (r = 0.966, P < 0.0001) doses. Conclusions Anthropometric estimates and medication dose recommendations provided by AiRDose strongly correlate with established techniques. Further study will establish the feasibility of using AiRDose to accurately obtain weight estimates and medication doses for pediatric patients in the prehospital setting.