EP338 - CONCORDANCE ON RESPIRATORY RATE EVALUATED THROUGH A VIDEO RECORDING IN CHILDREN WITH LOWER RESPIRATORY INFECTION (ID 1177)
Abstract
Backgrounds:
Respiratory distress is one of the most common causes of consultation at paediatric clinics. To establish its severity requires a clinician able to accurately determine the respiratory rate (RR).
VIBRA project aim is to develop a smartphone application to evaluate respiratory distress severity as efficiently as a trained paediatrician through a video of the patient’s chest and artificial intelligence algorithms.
To validate the application, a previous study to determine the concordance between two trained paediatricians measuring the RR of paediatric patients was conducted.
Methods
Patients with respiratory infections < 10 years are included. Medical information and 1-3 videos of each patient are recorded and shared on REDCap platform.
Two trained paediatricians evaluated and separately measured RR of videos included from November 2020-May 2021. Inter-observer concordance was measured with Bland-Althman method.
Results:
39 patients and 105 videos were included.
Median age was 1.1 (IQR: 0.09-3.71) years and 20(51.3%) were women. Main diagnosis were bronchiolitis (19;48.7%), bronchospasm (12;25.6%), and low respiratory tract infections (7;10.3%). RSV (12;30.7%), Rhinovirus (8;20.5%) and Parainfluenza (4;10.3%) were the most frequent detected etiologies. 28(66.7%) patients needed hospitalization and 22(56.4%), oxygentherapy.
In 93(88.5%) videos respiratory distress was seen. It was mild in 23(21.7%), moderate in 59(55.7%), and severe in 11(10.4%).
Mean RR was 41.3(±11.7) for observer1 and 42.7(±11.6) for observer2. Mean difference between the two observers was -1.06 (-1.74;-0.38), without statistical significance. This difference was not modified by low or high RR. Concordance limits were -5.22-3.1%, and 95.1% of the observations were concordant between the two observers.
Conclusions/Learning Points:
Results showed high concordance between two trained paediatricians for RR estimation in paediatric patients. This is the first step in order to validate a breath-rate measurement solution in the future.
