A premature infant born at 22 to 23 weeks’ gestation has skin so immature it functions almost like an open wound, losing water to the air at a rate that can dehydrate and dangerously chill the infant within minutes outside a tightly humidified incubator. To understand exactly how much water a given infant’s skin is actually losing, some NICU research teams use a specialized handheld probe called an evaporimeter, pressed directly against the infant’s skin, that measures transepidermal water loss — the literal rate of water evaporating from that patch of skin — rather than just reading the humidity of the surrounding air and assuming it reflects what’s happening at the skin’s surface.
Roughly a thousand miles away, in a commercial greenhouse full of tomatoes or strawberries, growers are trying to solve a version of the same underlying problem — how much water is actually leaving a living surface into the surrounding air — and, according to the industry’s own literature, still can’t do it directly at any practical scale. Two fields facing structurally the same measurement problem have each gotten good at a different half of the solution.
Scientific Foundation
Neonatal incubator design has converged on transepidermal water loss (TEWL) as the gold-standard way to understand water loss in extremely preterm infants; a study on infants born at 22 to 23 weeks measured evaporation rates directly from the skin using a dedicated device, finding that even carefully humidified incubators couldn’t fully prevent excessive water loss in the most immature infants. Yet the actual control systems that regulate incubator humidity remain comparatively simple: a 2021 engineering paper describing an automated humidity control system for neonatal incubators implements the control algorithm as an ON/OFF system with just four discrete levels of humidification power — a far cry from continuously modulated, individualized control driven by real-time water-loss feedback. A broader review of unresolved issues in incubator design similarly frames humidity and temperature control as an area still needing significant improvement.
Greenhouse and controlled-environment agriculture, meanwhile, has built its climate management around vapor pressure deficit, or VPD — the gap between the moisture the air can hold and what it currently holds, which directly governs how aggressively a plant’s stomata open and how much water it transpires. Modern greenhouse platforms now calculate VPD in real time across multiple zones, triggering automated alerts and corrective responses when conditions drift outside optimal ranges — a level of automated, continuously monitored environmental control that’s become standard in commercial agriculture. But growers’ own technical literature is candid about a persistent limitation: measuring the actual VPD at the leaf surface, where transpiration really happens, is “seldom practical,” because leaf temperatures vary wildly across a canopy depending on sun exposure, so most systems instead measure ambient air conditions near the canopy and use that as a proxy for what’s really happening at each individual leaf.
Cross-Domain Connection
The two fields have effectively split the same problem in half and solved opposite ends of it. Neonatal medicine has built and validated genuinely direct water-loss sensing — a probe that measures actual evaporation rate from a living surface, rather than inferring it from ambient conditions — but pairs that sensing capability with a fairly crude, discrete-level control system rather than continuous, individualized closed-loop regulation. Precision agriculture has built the opposite: sophisticated, continuously modulated, automated, multi-zone environmental control software, but by its own admission still approximates the thing it actually cares about — leaf-level transpiration — through ambient proxy measurements rather than direct sensing, because direct leaf-surface water-loss measurement isn’t practical at greenhouse scale.
There’s a real opportunity in each field borrowing the other’s stronger half. A miniaturized, non-contact or low-contact adaptation of NICU-style direct evaporation sensing, deployed on a subset of monitored plants within a greenhouse zone, could give agricultural VPD systems the direct transpiration ground-truth they currently lack and openly acknowledge approximating. In the other direction, NICU incubator design could benefit from adopting the kind of continuously modulated, algorithmically responsive control architecture greenhouse VPD systems already run commercially, replacing simple ON/OFF humidification stepping with the same kind of proportional, model-driven control loop that manages a multi-zone commercial grow operation.
What Remains Undemonstrated
Nobody has published research applying NICU-style direct transepidermal water-loss sensing technology to plant leaves, or greenhouse-style continuous VPD control architecture to neonatal incubators; the two literatures don’t reference each other. There are real physical differences that complicate a clean transfer: human skin and a plant leaf’s stomatal surface are different structures with different evaporation mechanics, and a sensing probe built for direct skin contact on a fragile premature infant would need substantial redesign to work non-invasively on a leaf without damaging it or interfering with photosynthesis. It’s also unproven that greenhouse-grade continuous control algorithms, built for larger, more thermally stable spaces, would translate cleanly to the much smaller, tightly enclosed volume of a neonatal incubator, where control dynamics behave differently.
Why It Matters
Both fields are, in the end, managing the same physics problem — regulating water loss from a delicate living surface via ambient humidity — for genuinely high-stakes reasons: preterm infant survival in one case, crop yield and disease prevention in the other. Neither field currently has the complete solution its counterpart has half-built. A direct exchange of NICU’s sensing precision and agriculture’s control sophistication wouldn’t require inventing new physics, just recognizing that two very different worlds have been quietly working on the same open question from opposite ends.
The Human Dimension
There’s something quietly moving about the idea that the same careful attention paid to a leaf trying not to wilt might, adapted and refined, help a NICU nurse know with more precision than an ON/OFF switch allows exactly how much water a one-pound infant’s skin is losing this hour. Both are forms of the same tenderness: trying to give something extraordinarily fragile exactly the environment it needs to survive, guided by better information than a best guess.
Sources:
1. “Incubator humidity and temperature control in infants born at 22–23 weeks’ gestation,” ScienceDirect: https://www.sciencedirect.com/science/article/pii/S0378378222000135
2. “Automated Humidity Control System for Neonatal Incubator,” ResearchGate: https://www.researchgate.net/publication/356545950_Automated_Humidity_Control_System_for_Neonatal_Incubator
3. “The infant incubator in the neonatal intensive care unit: unresolved issues and future developments,” PubMed: https://pubmed.ncbi.nlm.nih.gov/19591569/
4. “Ultimate Guide to Vapor Pressure Deficit,” Folio3 AgTech: https://agtech.folio3.com/blogs/ultimate-guide-to-vapor-pressure-deficit/
5. “Managing Vapor Pressure Deficit in Greenhouses,” Greenhouse Management: https://www.greenhouse-management.com/greenhouse_management/managing_vapor_pressure_deficit_greenhouse_crops/vapor_pressure_deficit_greenhouse_crops.htm
6. “What is the impact of VPD on crop development?” GLASE: https://glase.org/industry-news/what-is-the-impact-of-vpd-on-crop-development/
7. “Managing vapor pressures in controlled environment agriculture,” Greenhouse Product News: https://gpnmag.com/article/managing-vapor-pressures-in-controlled-environment-agriculture/
8. “Plant oriented control system based upon vapor pressure deficit data,” USPTO patent: https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/4858377
Idea originated at artificialideas.org. Article researched and written by Claude Sonnet 5. Published at artificialideas.org
Artificial Ideas 