Loading...
Outpatient evaluation of newborn jaundice relies on methods that are either unreliable (visual estimation) or expensive (transcutaneous and/or serum measurements). Smartphones may offer an inexpensive, widely available option for diagnosis based on digital imaging.
Researchers correlated bilirubin estimates derived from a novel smartphone app they developed with total serum bilirubin (TSB) in 530 newborns at seven U.S. sites. Six digital images of each newborn's sternum were taken (through a color calibration card) with a smartphone camera and sent via Internet for rapid analysis by a machine-learning algorithm. Sensitivity and specificity of app-derived bilirubin estimates for predicting high TSB were calculated using two common bilirubin decision rules: 1) ≥75th percentile on the Bhutani TSB nomogram to predict TSB in high-risk zone; 2) ≥13 mg/dL to predict TSB ≥17 mg/dL.
The correlation between app-derived bilirubin estimates and paired TSB was 0.91. Sensitivity and specificity for identifying high-risk TSB in the Bhutani nomogram were 85% and 75%, respectively, and for identifying TSB ≥17 mg/dL, 100% and 76%. These values were comparable with sensitivities and specificities of transcutaneous bilirubin estimates from a subset of study newborns and from other studies.
Taylor JA et al. Use of a smartphone app to assess neonatal jaundice. Pediatrics 2017 Aug 25; [e-pub]. (http://dx.doi.org/10.1542/peds.2017-0312)
Comment
A smartphone-based tool could greatly expand outpatient screening for jaundice and reduce the cost of determining which newborns need serum bilirubin measurements. For resource-poor areas of the world, where kernicterus continues to be a serious problem, the effect could be transformative. Though this study sample was racially diverse, further confirmation of the app's effectiveness is needed in larger populations across many types of settings.