P2.H158: Decision Accuracy of a Smartphone-based Serum Bilirubin Concentration Estimation App

Background: The Picterus Jaundice Pro smartphone application estimates neonates’ serum bilirubin concentrations using digital image color analysis and is approved for clinical use in the European Union. Because the American Academy of Pediatrics (AAP) recommends screening for hyperbilirubinemia in all newborns prior to discharge from the birth hospital, this smartphone-based approach might provide a new, non-invasive and low-cost alternative way to comply with the AAP guideline.

The objective of this study was to determine the accuracy of the Picterus Jaundice Pro smartphone app in a U.S. population, and specifically to quantify the risk that the smartphone estimate would fail to recommend phototherapy for a neonate who would have qualified for it if total serum bilirubin (TSB) screening were used.

Methods: Families of neonates born at 35 weeks’ gestation or greater, aged 24 to 192 hours, with a serum bilirubin level ordered by the care team were approached for consent. For those who consented, we obtained a non-invasive smartphone bilirubin estimate within 1 hour of drawing the TSB. This study only included neonates who had not received phototherapy. In addition to calculating common statistical measures of agreement, we quantified the risk that the smartphone estimate would fail to recommend phototherapy for a neonate who would have qualified for it if total serum bilirubin (TSB) screening were used.

Results: Two hundred twenty-two paired measurements were obtained from 100 neonates (TSB and smartphone estimate). The Pearson correlation coefficient between smartphone estimates and the TSBs was 0.90 (95% C.I.: 0.87-0.92). The mean bias of the smartphone estimates was -0.40 mg/dL (95% C.I.: -0.59-0.22 mg/dL). The smartphone app overestimated the TSB in 37.8% of the paired measurements and was within 2 mg/dL of the TSB in 79% of paired measurements. None of the 220 instances had a smartphone estimate indicating NO PHOTOTHERAPY, when the simultaneous TSB indicated PHOTOTHERAPY NEEDED. We estimate the risk of such an error occurring is between 0.0 and 2.1 per 100 measurements.

Conclusion: Smartphone-based bilirubin testing is an accurate method to estimate in-hospital, predischarge TSB. Using smartphone-based bilirubin estimates for screening at the birth hospital can identify those qualifying for phototherapy, with 0 of 220 paired measurements misidentifying a neonate as not needing phototherapy when by TSB they did.