Can a Smartphone Photo Tell If a Newborn Is Too Yellow?

What a Mexican study of the Picterus Jaundice Pro app found when it was tested against blood tests — and how well the old-fashioned eye exam still does

Researchers in Oaxaca, Mexico, tested a smartphone app that estimates a newborn's jaundice level from a photograph, comparing it with a laboratory blood test in 164 babies. The app was reasonably accurate for mild jaundice but tended to read too low when levels were high. A careful look at the baby by an experienced doctor did at least as well.

Why Newborn Jaundice Matters

Most newborn babies go a little yellow in their first week. The yellow colour comes from bilirubin, a natural pigment made when the body breaks down old red blood cells. Before birth, the mother's liver clears it away; afterwards the baby's own liver has to take over, and for a few days it is not quite up to speed. Bilirubin builds up, the skin and the whites of the eyes take on a yellow tint, and then — in the great majority of babies — the liver catches up and the colour fades. It is one of the most ordinary things that happens to newborns.

The reason clinicians take it seriously is a small minority of babies in whom bilirubin climbs much higher than usual. At very high levels the pigment can cross into the brain and damage it. The resulting condition, called kernicterus, can leave a child with lifelong hearing loss, difficulty controlling movement, and problems with learning — and at its worst it is fatal. What makes this so frustrating is that kernicterus is almost entirely preventable. If high bilirubin is found in time, the treatment is straightforward and safe: the baby lies under special blue lights, which change the bilirubin into a form the body can wash out quickly. The whole problem is one of finding the right babies in time [1].

Whether that happens depends enormously on where a baby is born. In wealthy countries, routine checks catch nearly all of these babies. In many low- and middle-income countries, kernicterus remains a major cause of newborn death and disability, not because the illness is different but because the tools to spot it are missing [2]. Worldwide estimates have put a number to this gap, attributing a substantial share of newborn deaths and of surviving disability to high bilirubin levels that were never caught [3].

How Clinicians Have Checked, Historically

For most of medical history the only tool was looking. In 1969 a doctor named Kramer noticed something useful: jaundice appears first on the face and spreads downwards over the body as bilirubin rises. He turned that observation into a simple five-zone score — yellow only on the face is zone one, yellow all the way to the soles and palms is zone five — which any clinician could use anywhere, with no equipment and no cost [4]. It is still used across the world today.

Looking, however, has real weaknesses. Later studies found that eyeballing jaundice was an unreliable way to screen for dangerously high bilirubin [5], and that estimates by eye could miss babies who genuinely needed treatment [6]. Lighting matters, skin tone matters, and experience matters a great deal. So hospitals moved to a handheld device, called a transcutaneous bilirubinometer, that is pressed briefly against the baby's forehead or chest and shines light into the skin to estimate bilirubin without a needle. Careful reviews have shown these devices agree well with blood tests, though not perfectly — well enough to decide who needs a blood test, not well enough to replace one [7]. One reason is that the device reads the pigment sitting in the skin, which is not quite the same as the pigment circulating in the blood [8]. The American Academy of Pediatrics' 2022 guidance builds this into a clear routine: use the skin device to screen, compare the result against charts based on the baby's exact age in hours, and confirm with a blood test whenever the number gets close to the treatment line [9].

The catch is price. As the Oaxaca researchers point out, a bilirubinometer typically costs between three and six thousand US dollars, which puts it out of reach of exactly the hospitals and clinics that need it most [1] — one strand of a much wider global inequality in newborn jaundice care that has been mapped in detail [10]. That is the gap that smartphones might fill — almost every clinic already has one. Several teams have tried: an app called BiliCam, tested in 530 newborns, matched blood tests closely [11]; an app called BiliScan was tested in a population of mixed ethnic backgrounds [12]; and another approach, photographing the whites of the eyes rather than the skin, was tested in Ghana [13]. The app in this study, Picterus Jaundice Pro (Picterus JP), was developed in Norway and first tested there [14], then refined and approved as a regulated medical device over several years [15]. A review pooling the evidence on smartphone jaundice apps concluded that the idea is promising, but that very little testing had been done in the places the tools are meant to serve [16]. That is exactly what this study set out to change.

What the Researchers Did

The study took place at a single hospital, Hospital General de Zona 1 (HGZ 1), part of the Mexican Institute of Social Security, in Oaxaca de Juárez, Mexico — a busy general hospital delivering around 2,000 babies a year in one of the country's poorer states. Between January 2023 and June 2024 the team enrolled 177 full-term babies aged between one and 14 days; 164 were included in the final analysis. The study was registered publicly before it began, under the reference NCT06276582.

The app works simply. A small printed colour card is laid on the baby's chest, and the phone automatically takes six photographs — three with the flash and three without. The card is the clever part: it gives the software a known set of colours in the same lighting as the baby, so it can correct for differences between phones and between rooms. The software then estimates the bilirubin level from how yellow the skin looks relative to that card.

Within an hour of the photographs, each baby had a small blood sample taken from the back of the hand and measured in the hospital laboratory. This was the yardstick everything else was judged against. Importantly, the babies were only ones who already needed a blood test for some other medical reason — no baby was pricked purely for research. The staff caring for the babies were not shown the app's estimate, and in fact the estimates were not even calculated until after the study finished, so the app could not influence anyone's decisions. At the same time, the study's doctors examined each baby and gave it a Kramer score, the traditional look-at-the-baby method.

What They Found

Of the 164 babies, 28 — about one in six — had bilirubin at or above the level the researchers defined as severe.

The app tracked the blood results reasonably well, but not perfectly. Set to raise a flag at its best-performing threshold, it correctly identified 85.7% of the babies who truly had severe jaundice, and correctly gave the all-clear to 80.1% of those who did not. In everyday terms: out of roughly every seven babies with dangerously high bilirubin, the app would have spotted six and missed one.

The traditional eye examination did better on every measure. The study's experienced doctors, using the Kramer score alone, correctly identified 92.9% of the severe cases and correctly cleared 85.6% of the rest. This is worth sitting with. The point is not that phones are useless, but that a well-trained clinician looking carefully at a baby is a genuinely good test — and that training people is as valuable an investment as buying technology.

The most important finding concerns where the app went wrong. For babies with lower bilirubin levels, the app was almost exactly right on average. For babies above the severe threshold, it read too low — by about 90 units on average, a large gap. That is the worst possible place for an error, because it is the babies with the highest levels who most urgently need treatment. An app that quietly underestimates the sickest babies could offer false reassurance precisely when it matters most. The researchers were candid about this, and it is the main thing they say has to be fixed.

What This Means for Families, and What Comes Next

If you are a parent, the practical message is that a phone photograph is a check, not a verdict. A tool like this could be genuinely valuable in a clinic that has no other way of measuring bilirubin, helping staff decide which babies need to travel for a blood test. It is not a reason to skip a blood test in a baby who looks unwell. Trust what you see: deepening yellow spreading down to the legs and feet, a baby who is unusually sleepy, feeding poorly, or crying in a high-pitched way, always deserves to be seen — whatever any app says.

The researchers are clear about what needs to happen next. The software needs to be corrected so it stops reading low at high bilirubin levels. It needs testing in babies with darker skin, because this study could not answer that question — the skin-tone measurements the team collected turned out not to be reliable enough to analyse, and most of the babies had lighter skin. The app also currently needs a good internet connection and a reasonably modern phone, both of which cannot be assumed in the settings it is designed for. And the team recommends better training in visual assessment for staff everywhere. The same group has since carried out a further study looking at how the technology works when actually put into use in this setting [17].

One thing readers should know in weighing all this: two of the study's authors work for and own shares in the company that makes the app, and the company supplied the phones and cards. The researchers disclosed this openly — and it is notable that the study still reported the free, low-tech method coming out ahead.

References

  1. Jiménez-Díaz G, Gierman LM, Keitsch M, et al. Validation of a Mobile Health Device for Neonatal Jaundice Screening: A Cross-Sectional Study in a Resource-Limited Setting in Mexico. Sage Open Pediatrics. 2025;12. doi:10.1177/30502225251320544 ↩
  2. Olusanya BO, Ogunlesi TA, Slusher TM. Why is kernicterus still a major cause of death and disability in low-income and middle-income countries? Arch Dis Child. 2014;99(12):1117–1121. doi:10.1136/archdischild-2013-305506 ↩
  3. Bhutani VK, Zipursky A, Blencowe H, et al. Neonatal hyperbilirubinemia and Rhesus disease of the newborn: incidence and impairment estimates for 2010 at regional and global levels. Pediatr Res. 2013;74(Suppl 1):86–100. doi:10.1038/pr.2013.208 ↩
  4. Kramer LI. Advancement of dermal icterus in the jaundiced newborn. Am J Dis Child. 1969;118(3):454–458. doi:10.1001/archpedi.1969.02100040456007 ↩
  5. Riskin A, Tamir A, Kugelman A, Hemo M, Bader D. Is visual assessment of jaundice reliable as a screening tool to detect significant neonatal hyperbilirubinemia? J Pediatr. 2008;152(6):782–787. doi:10.1016/j.jpeds.2007.11.003 ↩
  6. Keren R, Tremont K, Luan X, Cnaan A. Visual assessment of jaundice in term and late preterm infants. Arch Dis Child Fetal Neonatal Ed. 2009;94(5):F317–F322. doi:10.1136/adc.2008.150714 ↩
  7. Okwundu CI, Olowoyeye A, Uthman OA, et al. Transcutaneous bilirubinometry versus total serum bilirubin measurement for newborns. Cochrane Database Syst Rev. 2023;(5):CD012660. doi:10.1002/14651858.CD012660.pub2 ↩
  8. Bosschaart N, Kok JH, Newsum AM, et al. Limitations and opportunities of transcutaneous bilirubin measurements. Pediatrics. 2012;129(4):689–694. doi:10.1542/peds.2011-2586 ↩
  9. Kemper AR, Newman TB, Slaughter JL, et al. Clinical practice guideline revision: management of hyperbilirubinemia in the newborn infant 35 or more weeks of gestation. Pediatrics. 2022;150(3):e2022058859. doi:10.1542/peds.2022-058859 ↩
  10. Olusanya BO, Kaplan M, Hansen TWR. Neonatal hyperbilirubinaemia: a global perspective. Lancet Child Adolesc Health. 2018;2(8):610–620. doi:10.1016/S2352-4642(18)30139-130139-1) ↩
  11. Taylor JA, Stout JW, de Greef L, et al. Use of a smartphone app to assess neonatal jaundice. Pediatrics. 2017;140(3):e20170312. doi:10.1542/peds.2017-0312 ↩
  12. Ngeow AJH, Tan MG, Dong X, et al. Validation of a smartphone-based screening tool (Biliscan) for neonatal jaundice in a multi-ethnic neonatal population. J Paediatr Child Health. 2023;59(2):288–297. doi:10.1111/jpc.16287 ↩
  13. Enweronu-Laryea C, Leung T, Outlaw F, et al. Validating a sclera-based smartphone application for screening jaundiced newborns in Ghana. Pediatrics. 2022;150(1):e2021053600. doi:10.1542/peds.2021-053600 ↩
  14. Aune A, Vartdal G, Bergseng H, Randeberg LL, Darj E. Bilirubin estimates from smartphone images of newborn infants' skin correlated highly to serum bilirubin levels. Acta Paediatr. 2020;109(12):2532–2538. doi:10.1111/apa.15287 ↩
  15. Aune A, Vartdal G, Jimenez Diaz G, Gierman LM, Bergseng H, Darj E. Iterative development, validation, and certification of a smartphone system to assess neonatal jaundice: development and usability study. JMIR Pediatr Parent. 2023;6:e40463. doi:10.2196/40463 ↩
  16. Hegde D, Rath C, Amarasekara S, Saraswati C, Patole S, Rao S. Performance of smartphone application to accurately quantify hyperbilirubinemia in neonates: a systematic review with meta-analysis. Eur J Pediatr. 2023;182(9):3957–3971. doi:10.1007/s00431-023-05073-2 ↩
  17. Jiménez-Díaz G, Elizarrarás-Rivas J, Keitsch M, Gierman LM, Marcuzzi A, Infanti JJ. Advancing neonatal jaundice screening with mHealth: a mixed-methods randomised controlled feasibility study in a resource-limited setting in Mexico. BMJ Paediatr Open. 2026:e003624. doi:10.1136/bmjpo-2025-003624 ↩