Watching a Baby Breathe Without Stickers or Wires: How Close Are Cameras to Helping in Neonatal Intensive Care?

What a 24-hour camera study in Cambridge, a smartphone video study in Brazil and a review of 60 earlier studies tell families about contactless breathing monitors

Babies in neonatal intensive care wear sticky sensors and wires so that nurses can track their breathing around the clock, but those sensors can mark delicate skin and get in the way of cuddles. New research shows that a camera placed outside the incubator can count a premature baby's breaths to within about 1.5 breaths a minute of the hospital monitor, on average, when both get a clear reading. It is not yet ready to replace the monitor, especially for spotting pauses in breathing, but it is a promising step towards gentler care.

Why breathing is watched so closely, and why the wires matter

When a baby is born early or is unwell, the neonatal intensive care unit (NICU) team keeps a constant eye on heart rate, breathing and oxygen levels. Doctors have learned that patterns in these numbers over hours and days can give early warning of serious problems such as infection [1]. Today, breathing is usually measured through small sticky pads on the chest, called electrodes, that also record the heartbeat, along with other sensors taped or strapped to the baby [2][3]. The glue on these pads can tear and irritate the skin and cause pain, adding to discomfort and the risk of infection [2]. That matters most for a very premature baby, whose skin is still fragile and immature for the first few weeks of life and is easily damaged [4]. Parents and nurses in one British unit described how the tangle of wires made it harder to hold and bond with their baby and slowed down everyday care [5]. Even the usual method is not perfect: researchers have shown it can struggle to detect some pauses in breathing in premature babies [6]. So doctors, nurses and engineers have been asking a simple question for years: could we keep the same close watch on a baby's breathing while touching them less? The answer matters for comfort, for skin health and for the precious time parents spend holding their baby.

How researchers have tried to measure breathing without touching

For more than ten years, engineers and doctors have tried to measure breathing in other ways. In Oxford, a team filmed premature babies in their incubators with ordinary cameras for days at a time [7]. In a later study of 30 babies, their system's count was on average about 3.5 breaths per minute different from the monitor's, but only when the baby was in view and nobody was caring for them [8]. Other teams tried heat-sensing cameras [9], depth cameras that measure distance and can detect chest movements of around one millimetre [10], small radar devices [11][12], and wireless sensors that are soft and thin like a sticker but have no cables [13]. Most of these early studies filmed only a handful of babies, often for minutes rather than hours, and usually chose moments when the baby was lying still [2].

A review that looked at 60 studies

A team at McGill University and the Montreal Children's Hospital in Canada gathered every study they could find, published between 2014 and 2024, that tested a camera, radar or wireless sensor on babies in neonatal intensive care [2]. They found 60. A typical study included just 10 babies and recorded each one for about 19 minutes. Most measured only one thing, usually breathing rate, and only checked how accurate the device was. Only one study out of 60 checked whether the device was safe for the skin. The reviewers concluded that the technology is promising but not ready for everyday use, and that future studies need longer recordings, a wider range of babies, and checks on safety and practicality. This review was finished before the two new studies below were published, and it describes exactly the gaps they set out to fill.

A camera above the incubator for a whole day

At Addenbrooke's Hospital in Cambridge, UK, researchers placed a special camera just outside the incubators of 20 premature babies and recorded each baby for about 24 hours [3]. The camera measures depth as well as colour, so it can "see" the rise and fall of the chest even in the dark. Nothing touched the baby, and nothing about the babies' care was changed. On average the babies had been born after about 27 weeks of pregnancy and weighed 950 grams at birth; all but one needed help with breathing, most of them through soft tubes in the nose. A computer program found the baby in the picture and counted breaths automatically, and the results were compared with the standard hospital monitor.

What the camera got right, and where it struggled

When both the camera and the monitor gave a clear reading, they usually agreed closely. On average, the camera's count was about 1.5 breaths per minute different from the monitor's. Almost all the readings in this study were faster than 35 breaths a minute, so that is a small difference. Each device gave a usable reading a little more than half the time. Importantly, they tended to have trouble at different moments, so using both together meant a usable reading about three-quarters of the time.

The camera found things harder in situations parents will recognise. When a blanket covered the baby's chest or the baby was wriggling, it gave usable readings less often. It was less accurate in the smallest babies, under 1,000 grams, in babies on a breathing support called CPAP (a gentle flow of air pressure through the nose), and when breathing was slow. Slow breathing happened very rarely during the recordings, and the monitor itself is not very reliable at slow rates, so the researchers could not judge this properly.

A smartphone and computer program in Brazil

In a hospital in Brazil, another team tried something simpler: short videos, 2 to 5 minutes long, filmed with ordinary smartphones placed about 30 cm from babies lying still in a nappy [14]. The babies were close to their due date and breathing normal air without help. Artificial intelligence (computer software that learns patterns from examples) found the chest and tummy in the picture and counted breaths, pausing whenever the baby moved. In 23 recordings, the program's count was on average 2.1 breaths per minute different from that of an experienced health professional counting by eye. The team says this shows the idea works technically but must now be tested in sicker babies and in more hospitals.

Why the two results are not a contest

It is natural to wonder which approach is "better", but these studies cannot be compared like that. The Cambridge camera worked on tiny premature babies in incubators through day and night, while the smartphone study looked at bigger, healthier babies for a few calm minutes. The Cambridge team showed how much conditions matter: when they looked only at quiet hours with the baby uncovered, their error fell to about half a breath per minute [3]. That is one reason the review team warned that studies which only use calm moments can make devices look better than they will be in real life [2].

What this means for your baby today

If your baby is in neonatal intensive care now, they will still be monitored with the usual sensors, and that is the right thing. Neither of the two new studies tested whether a camera can reliably spot pauses in breathing, called apnoea, and the Cambridge team says this should be the focus of future work [3]. You can read more about this in our article on apnoea of prematurity. If you are asked to take part in a study like these, it is reasonable to ask how the video will be stored and who can see it; the Cambridge team suggests video should stay in the hospital and not be kept once the breathing signal has been extracted [3].

Questions researchers are working on next

The researchers agree on what comes next. They want to test cameras and wireless sensors in many more babies, in several hospitals, including the smallest babies and those on breathing support [2][3]. They need to check that devices work equally well for babies of every skin colour; most babies in the Cambridge study were White, and the review lists skin colour as an unanswered question for camera-based methods. Our article on pulse oximeters and skin colour explains why this matters. Researchers also want to test wireless sensors that avoid cables altogether; one Canadian team has planned to monitor babies with such a system for up to four days in a row [15]. Others hope cameras may one day help nurses spot signs that a baby is working hard to breathe [14].

Some of the researchers involved also work on developing wireless monitoring devices, which they declared openly in their papers. The review team points out that such declarations matter in this field because the devices have a high potential to be sold commercially [2], and this is one reason independent studies in other hospitals will be important.

The bottom line for families

Cameras that watch a baby breathe without touching them are getting closer to real use. In one hospital, a camera outside the incubator matched the monitor closely when it had a good view, and together the two gave more continuous information than either alone. But blankets, movement, very small size and breathing support all made the camera's job harder, and it has not yet been tested for spotting breathing pauses. For now, these are research tools that may one day mean fewer stickers and wires, and more freedom to hold your baby close, such as during skin-to-skin care.

References

  1. Kumar N, Akangire G, Sullivan B, Fairchild K, Sampath V. Continuous vital sign analysis for predicting and preventing neonatal diseases in the twenty-first century: big data to the forefront. Pediatr Res. 2020;87(2):210-220. doi:10.1038/s41390-019-0527-0 ↩
  2. Senechal E, Maximov A, Jeanne E, et al. Next generation of non-contact and wireless vital sign monitoring technology in the neonatal intensive care unit: a systematic review. Pediatr Res. 2026;99(7):2490-2498. doi:10.1038/s41390-025-04469-0 ↩
  3. Grafton A, Thomson L, Beardsall K, Lasenby J. Camera-based respiratory rate monitoring in unmodified neonatal intensive care - a cross-sectional study in a UK hospital. EClinicalMedicine. 2026;99:104175. doi:10.1016/j.eclinm.2026.104175 ↩
  4. Oranges T, Dini V, Romanelli M. Skin Physiology of the Neonate and Infant: Clinical Implications. Adv Wound Care (New Rochelle). 2015;4(10):587-595. doi:10.1089/wound.2015.0642 ↩
  5. Bonner O, Beardsall K, Crilly N, Lasenby J. 'There were more wires than him': the potential for wireless patient monitoring in neonatal intensive care. BMJ Innov. 2017;3(1):12-18. doi:10.1136/bmjinnov-2016-000145 ↩
  6. Lim K, Eastwood-Sutherland C, Marshall AP, Gale TJ, Dargaville PA. Limitations of thoracic impedance monitoring for central apnoea detection in preterm infants. Acta Paediatr. 2021;110(9):2550-2552. doi:10.1111/apa.15888 ↩
  7. Villarroel M, Guazzi A, Jorge J, et al. Continuous non-contact vital sign monitoring in neonatal intensive care unit. Healthc Technol Lett. 2014;1(3):87-91. doi:10.1049/htl.2014.0077 ↩
  8. Villarroel M, Chaichulee S, Jorge J, et al. Non-contact physiological monitoring of preterm infants in the Neonatal Intensive Care Unit. NPJ Digit Med. 2019;2:128. doi:10.1038/s41746-019-0199-5 ↩
  9. Pereira CB, Yu X, Goos T, et al. Noncontact Monitoring of Respiratory Rate in Newborn Infants Using Thermal Imaging. IEEE Trans Biomed Eng. 2019;66(4):1105-1114. doi:10.1109/TBME.2018.2866878 ↩
  10. Ottaviani V, Veneroni C, Dellaca' RL, Lavizzari A, Mosca F, Zannin E. Contactless Monitoring of Breathing Pattern and Thoracoabdominal Asynchronies in Preterm Infants Using Depth Cameras: A Feasibility Study. IEEE J Transl Eng Health Med. 2022;10:4900708. doi:10.1109/JTEHM.2022.3159997 ↩
  11. Kim JD, Lee WH, Lee Y, et al. Non-contact respiration monitoring using impulse radio ultrawideband radar in neonates. R Soc Open Sci. 2019;6(6):190149. doi:10.1098/rsos.190149 ↩
  12. Beltrão G, Stutz R, Hornberger F, et al. Contactless radar-based breathing monitoring of premature infants in the neonatal intensive care unit. Sci Rep. 2022;12(1):5150. doi:10.1038/s41598-022-08836-3 ↩
  13. Chung HU, Kim BH, Lee JY, et al. Binodal, wireless epidermal electronic systems with in-sensor analytics for neonatal intensive care. Science. 2019;363(6430):eaau0780. doi:10.1126/science.aau0780 ↩
  14. Ribeiro SNS, Fernandes AER, Vargas MERR, et al. Video-based artificial intelligence for automated neonatal respiratory monitoring. Eur J Pediatr. 2026;185(8):617. doi:10.1007/s00431-026-07267-w ↩
  15. Senechal E, Radeschi D, Tao L, et al. The use of wireless sensors in the neonatal intensive care unit: a study protocol. PeerJ. 2023;11:e15578. doi:10.7717/peerj.15578 ↩