JMIR Publications

ABSTRACT

Background:

The early access to prenatal care and high-cost technologies for pregnancy dating challenge the early neonatal risks assessment at birth in resource-constrained settings. To overcome the absence or low accuracy of postnatal gestational age, we developed a frugal innovation based on the photobiological properties of the newborn's skin and predictive models.

Objective:

This study aims to validate the photobiological model of skin maturity adjusted to the clinical data to promptly detect gestational age and determine its accuracy in detecting prematurity.

Methods:

A multicenter single-blinding and single-arm clinical trial intention-to-diagnosis evaluated the accuracy of a novel device to detect gestational age and preterm newborns. The first-trimester ultrasound (US), a second comparator US, and the last menstrual period (LMP) data from antenatal reports were the references for gestational age at birth. A portable multiband reflectance photometer assessed 781 newborns’ skin maturity and used machine learning models to predict gestational age, adjusted to birth weight and antenatal corticosteroid therapy exposure.

Results:

As the primary outcome, the predicted gestational by the new test had high agreement with the reference gestational age calculated with the intraclass correlation coefficient (0.970 [95%CI: 0.965, 0.974]) similar values to the comparator-US and better than the comparator-LMP gestational ages. As secondary outcomes, the new test achieved 97.7% (95%CI: 96.5%, 98.6%) agreement with the reference gestational age within one-week error. This value surpassed those of comparator-US (91.3% [95%CI: 89.2%, 93.1%]), and of comparator-LMP gestational ages (64.1% [60.7% to 67.5%]). Bland-Altman limits of the new test were -7.1 to 4.7 days. Prematurity discrimination with the novel device had the area under the receiver operating characteristic curve (AUROC) (0.998 [95%CI: 0.997, 1.000]), similar to comparator-US (0.996 [95% CI: 0.993, 0.999)]; and superior to comparator-LMP gestational ages (0.957 [95%CI:0.941, 0.974]). In newborns with absent or unreliable LMP (n=451), the intent-to-discriminate analysis showed correct classifications with the new test of 96.5% (95%CI: 94.3%, 98.0%), while with the comparator-LMP gestational age was 69.6% (95% CI: 65.3%, 73.7%).

Conclusions:

The assessment of the newborn's skin maturity adjusted by learning models promises accurate pregnancy dating at birth even without the antenatal ultrasound reference. Clinical Trial: WHO’s International Clinical Trial Platform - Brazilian Clinical Trials Registry RBR-3f5bm5.