ADVANCES
20 Scientific American, June 2022Alana Chin/University of California, DavisT E C HVitamin Map
AI can pinpoint nutrient deficiencies from space
Micronutrient deficiencies afflict more than two billion people
worldwide, including 340 million children. This lack of vitamins
and minerals can have serious health consequences. But diagnos-
ing deficiencies early enough for effective treatment requires
expensive, time-consuming blood draws and laboratory tests.
New research provides a more efficient approach. Computer
scientist Elizabeth Bondi and her colleagues at Harvard Univer-
sity used publicly available satellite data and artificial intelli-
gence to reliably pinpoint geographical areas where populations
are at high risk of micronutrient deficiencies. This analysis could
potentially pave the way for early public health interventions.
Existing AI systems can use satellite data to predict localized
food security issues, but they typically rely on directly observable
features. For example, agricultural productivity can
be estimated from views of vegetation. Micro-
nutrient availability is harder to calculate.
After seeing research showing that areas
near forests tend to have better dietary
diversity, Bondi and her colleagues
were inspired to identify lesser-known
markers for potential malnourishment.
Their work shows that combining data
such as vegetation cover, weather and
water presence can suggest where popu-
lations will lack iron, vitamin B 12 or vitamin A.
The team examined raw satellite measure-
ments and consulted with local public health officials, then used AI
to sift through the data and pinpoint key features. For instance, a
food market, inferred based on roads and buildings visible, was vital
for predicting a community’s risk level. The researchers then linked
these features to specific nutrients lacking in four regions’ popula-
tions across Madagascar. They used real-world biomarker data
(blood samples tested in labs) to train and test their AI program.
Predictions of regional-level micronutrient deficiency in popula-
tions outside the training data sets met, and sometimes exceeded,
the accuracy of estimates based on surveys administered by local
public health officials. “Our work showcases a method that allows
for identification and targeting of vulnerable populations for nutri-
tional support that may supplement ... expensive and invasive pro-
cedures,” Bondi says. The study was detailed at the Association for
the Advancement of Artificial Intelligence’s 2022 virtual meeting.
“This is a novel contribution that highlights AI’s potential to
advance public health,” says Emory University epidemiologist
Christine Ekenga, who was not involved with the study. Collect-
ing health data in low-resource settings can be difficult because
of cost and infrastructure constraints, she adds, and “the authors
have validated a method that can overcome these challenges.”
The researchers aim to develop a software application that
extends this analysis to other countries that have public satellite
data. “We hope that this application could allow public health
officials to interact with the insights our system can provide and
help to inform interventions,” Bondi says. — Rachel BerkowitzE C O L O G YSmart Leaves
Thirsty redwoods pull water out of thin air
Coastal California’s redwood forests—with their lush ferns,
towering trees and damp petrichor scent—might not seem to
want for water, but they do face dry summers. To survive them,
the trees, Sequoia sempervirens, grow specialized shoots with
leaves that scrape moisture from the air.
Many plants (including redwoods) are known to drink through
their leaves, but “no one ever really figured out how the water
gets in there,” says ecologist Alana Chin, now at ETH Zürich.
Exposing leaves to moisture has costs: even a thin film of water
can block the flow of carbon dioxide into leaf openings called
stomata, hindering photosynthesis.
To see how the trees solve this dilemma, Chin and her col-
leagues climbed redwoods in various climate zones and snipped
twig samples. Back in the laboratory, they generated fog with a
humidifier and measured how much water these leafy shoots
absorbed. They also examined leaf surfaces and cross sections,
then modeled water movement to see which traits affect uptake.
Their analysis, published in the American Journal of Botany ,
revealed two distinct redwood shoot types. Resembling asparagus
stalks with leaves bunched close to the twig, “axial” shoots make
up a small portion of the canopy but absorb water at about four
times the rate of ordinary-looking “peripheral” shoots. The team
estimated a tall redwood absorbs up to 13 gallons of water in the
hour after it gets wet. Meanwhile peripheral leaves power photo-
synthesis with dense stomata and waxy, water-repellent coatings.
The study found that redwoods in drier, southern areas have
more axial shoots that are located higher up than on northern trees,
which helps the former pull extra water from summer fog and light
rain. Other tree species may have similarly specialized shoots;
pines, for example, have two types that might be analogous to
those on redwoods, Chin says. Such versatility could be important
in the context of climate change, notes Wake Forest University
ecologist Carter Berry, who was not involved in the study. “In a
drier world,” he says, “the ability to subsidize your water source
with water from the air becomes more important.” — Ula ChrobakIllustrations by Thomas Fuchs