What can a Giraffe Teach Us About Female Biology?

The hidden physiology that could reshape medicine, if our systems learned to see it

A pregnant giraffe routinely sustains a blood pressure around 300 over 180.

In a pregnant woman, those numbers signal a hypertensive emergency that can kill mother and child within hours. The giraffe tolerates these extreme pressures safely as a normal routine.

Its left ventricle thickens under the lifelong load of pumping blood up a two-meter neck, yet it does not stiffen and fail the way a human heart often does under chronic pressure. The giraffe appears to have evolved, over roughly eleven and a half million years, resistance to a version of the problem human medicine still struggles to treat: pressure overload without the same progression to heart failure with preserved ejection fraction, a condition disproportionately common in older women and still difficult to treat.

This is the work of Barbara Natterson-Horowitz, a cardiologist and evolutionary biologist (whom I had the pleasure of hearing speak recently at the fabulous Wyss Institute) and who asks a question most of medicine never thought to pose: what if other species have already evolved answers to the problems killing human patients, and the only reason no one noticed is that human medicine assumed it had nothing to learn from a zoo?

The tree of life is a vast, unexamined library of solved physiology, and the female branches of that tree, precisely because medicine ignored female biology for so long, are among the least-read shelves in the building.

The giraffe did not “solve” human heart failure. But it evolved resistance to a pressure-load problem that human medicine still struggles to manage. No one looked closely enough for long enough, because medicine did not think a giraffe had anything to teach it about a woman’s heart.

The deeper point is not the giraffe specifically. It is the category of error the giraffe exposes.

Treating female biology as a complicated deviation from a male norm did not only create injustice but a vast scientific blind spot.

It produced a scientific blind spot the size of half the species, in which whole mechanisms of resilience, defense, and repair went uninvestigated despite being visible all along. Once you start looking, the pattern repeats across systems. The most consequential examples are often not rare diseases or edge cases but ordinary features of female physiology that medicine observed, compartmentalized, and then failed to follow to their broader implications.

The immune system offers perhaps the clearest example:

I. One of the most sophisticated acts of immune control in human biology happens routinely, in millions of bodies: pregnancy.

During pregnancy, a woman carries a fetus that is genetically half foreign. The fetus expresses paternal proteins that the mother’s immune system could, in another context, recognize as non-self. And yet, for nine months, the maternal immune system usually does not reject it.

This is extraordinary. In transplantation, foreign tissue often requires immunosuppression to prevent rejection. In pregnancy, the body achieves something far more elegant: it creates tolerance locally, without shutting down the entire immune system.

Peter Medawar described this as one of the central paradoxes of immunology in 1953. It is still not fully solved. But it may also be one of the richest therapeutic clues in medicine.

A major part of the machinery appears to run through regulatory T cells, or Tregs, which act like the immune system’s brakes. During pregnancy, these cells expand and help create tolerance at the maternal-fetal interface, where the mother’s body and the fetus are in constant biological conversation.

This insight matters in two opposite directions.

First, pregnancy shows that the body can tolerate genetically distinct tissue without collapsing its broader immune defenses. That makes maternal-fetal tolerance a powerful natural model for autoimmune disease and transplantation. If we could learn how pregnancy induces targeted tolerance, we might be able to treat autoimmune disease or prevent transplant rejection without relying on blunt, lifelong immune suppression.

Second, pregnancy helps explain one of cancer’s central tricks. Tumors often exploit immune-tolerance mechanisms, including Treg-mediated suppression, to hide from immune attack. In pregnancy, tolerance is controlled and purposeful. In cancer, similar brakes can be captured by disease.

The natural experiment is visible in patients themselves. Several autoimmune diseases, including multiple sclerosis and some cases of rheumatoid arthritis and psoriasis, often improve during pregnancy and then flare again after delivery. Pregnancy becomes a biological on-off switch, running on a nine-month clock.

That is what makes the field so powerful. Maternal-fetal tolerance and tumor immune escape are related problems with the sign flipped. Pregnancy requires tolerance. Cancer exploits tolerance. Autoimmune disease and transplant rejection may require us to induce tolerance deliberately.

Studying how a fetus is tolerated can therefore help answer three enormous medical questions: how tumors hide, how immune brakes can be released to make cancer visible again, and how tolerance might be induced in the opposite direction for autoimmune disease and transplantation.

This is the holy grail for fields that still rely too often on broad immune suppression. Too little immune tolerance can contribute to pregnancy complications such as miscarriage and preeclampsia. Too much local immune tolerance can help cancers evade attack. The relationship is not universal, but the pattern is profound.

One mechanism, studied in pregnant women, opens two of the largest therapeutic frontiers in medicine: teaching the immune system when to stand down, and when to fight.

II. What if we viewed female biology as a discovery platform ? What differences are contained within the chromosomes ?

Every female cell carries two X chromosomes. Male cells usually carry one X and one Y. To prevent female cells from producing a double dose of X-linked genes, one of the two X chromosomes is largely silenced through a process called X-inactivation.

For a long time, scientists assumed this solved the dosage problem. It does not.

A meaningful share of X-linked genes escape inactivation and remain active on both chromosomes. This means that some female cells can express higher levels of certain genes than male cells. What once looked like a minor genetic exception is now understood as a fundamental biological difference, with consequences for immunity, metabolism, and disease.

One of the clearest examples is TLR7, a gene involved in the immune system’s ability to detect viruses. Because TLR7 can escape X-inactivation, some female immune cells may express it from both X chromosomes.

That extra activity can be protective. Stronger TLR7 signaling may help the immune system mount a faster antiviral response. Rare loss-of-function mutations in TLR7 have been linked to severe COVID-19 in young men, suggesting that adequate TLR7 activity is important for early immune defense.

But the same pathway can also become dangerous. Too much TLR7 signaling can push the immune system toward autoimmunity. In mice, increased TLR7 activity is sufficient to drive lupus-like disease, and in humans, lupus occurs far more often in women than in men.

The same mechanism, then, can cut in two directions. More immune vigilance may improve defense against infection. But too much immune vigilance may increase the risk of autoimmune disease.

This is the deeper point: sex differences in disease are not only about hormones.

Researchers have shown this using the Four Core Genotypes mouse model, which separates the effects of sex chromosomes from the effects of reproductive organs and hormones. In this model, scientists can study mice that are XX with testes or XY with ovaries. This allows them to ask whether a trait is driven by chromosomes themselves, by hormones, or by both.

The results showed that chromosomes matter in their own right. In some experiments, mice with two X chromosomes accumulated more body fat and showed greater metabolic dysfunction on high-fat diets, even when hormonal effects were controlled. This demonstrated that sex differences in metabolism cannot always be explained by estrogen, testosterone, ovaries, or testes alone.

That distinction becomes especially important later in life. Women’s health is often discussed almost entirely through the lens of hormones, especially menopause. Hormones matter enormously. But they are not the whole story. Chromosome dosage may also shape disease risk, and those effects may become more visible when hormonal patterns change.

Human evidence points in the same direction. Men with Klinefelter syndrome, who carry an extra X chromosome and have a 47,XXY karyotype, develop lupus at rates much closer to 46,XX women than to typical 46,XY men. That suggests the number of X chromosomes itself contributes to autoimmune risk, independent of whether someone has ovaries.

The broader lesson is simple but profound: female biology is not just male biology plus hormones.

It contains distinct genetic, immune, and metabolic architectures. These architectures are not edge cases. They are clues. They can help explain why some diseases appear more often in women, why immune responses differ by sex, and why therapies developed around male biology may miss important mechanisms.

By studying female biology directly, medicine does not merely correct an oversight. It gains access to new biological control points: ways to tune immunity, metabolism, inflammation, and disease susceptibility with far greater precision.

III. Lactation: a window into osteoporosis-prevention?

Osteoporosis is often framed as a story of female vulnerability: women lose bone after menopause, suffer fractures, and face a long decline. But that framing overlooks one of the most extraordinary examples of controlled regeneration in human biology. During breastfeeding, a mother’s skeleton releases large amounts of calcium into her milk. In the spine, bone mineral density can fall by five to ten percent over just a few months, a level of bone loss that would be alarming in almost any other setting. Yet after weaning, much of that bone is rebuilt, often within a year. The body intentionally breaks down part of its own skeleton and then restores it.

This matters because rebuilding bone is one of the hardest problems in osteoporosis medicine. Preventing further loss is possible; restoring lost bone is much more difficult. Lactation provides a natural example of a biological system that does exactly that. Bone density falls rapidly during breastfeeding through a hormonally programmed process that is largely independent of how much calcium a woman consumes. Then, once breastfeeding ends, bone formation accelerates and recovery begins.

The most intriguing part is that scientists still do not fully understand what triggers this rebuilding phase. Researchers have mapped much of the bone-loss process to low estradiol levels and high levels of parathyroid-hormone-related protein (PTHrP). But studies in genetically modified animals show that the recovery phase does not depend on PTHrP, calcitriol, or parathyroid hormone (PTH) in the straightforward way researchers expected. In other words, the body carries out one of the most effective bone-rebuilding programs in human physiology, on a predictable schedule, and medicine still cannot fully explain how it works. That makes lactation not a solved problem, but a valuable source of clues for future osteoporosis therapies.

Comparative biology reinforces the lesson. Hibernating bears spend months almost completely inactive, a condition that should cause severe bone loss. Instead, they suppress bone breakdown and emerge with their skeletons largely intact. Together, lactation and hibernation point to the same idea: biological systems already exist that can preserve or rebuild bone. What medicine often sees as inevitable decline may, in another context, be a controlled and reversible process.

IV. The placenta: controlled invasion without metastasis

The placenta is the one organ humans grow and discard on demand, and it does something no other healthy tissue is normally permitted to do. It invades. Trophoblast cells burrow into the uterine wall, break down and remodel the mother’s spiral arteries, divert her blood supply to the fetus, and evade her immune system, even borrowing the molecular tools of blood-vessel cells, VE-cadherin among them, to build vascular channels of their own. Proliferation, invasion, immune modulation, angiogenesis: that list is cancer-like.

But the placenta is not a tumor.

In a normal pregnancy, trophoblast invasion is spatially and temporally finely controlled: the invading cells advance, remodel, and halt at the inner third of the uterine muscle, on schedule. The same behaviors that make a tumor lethal are executed by the placenta under tight regulation and then switched off. These processes respond to a controlled program in trophoblasts, while in cancer cells this regulation is lost. When the brake fails toward excess, when invasion goes too deep, the result is a dangerous obstetric condition, placenta accreta.

A tumor behaves like invasion with its brakes cut. The placenta shows that invasion can be programmed, bounded, and stopped. Understanding how it knows when to halt is a live lead on the hardest problem in cancer: metastasis.

This is why placental biology is increasingly read as a roadmap for cancer, and the link is now argued to be more than analogy. Comparative work shows that species whose placentas evolved to be non-invasive, cows among them, also have stroma that resists cancer-cell invasion, suggesting the two phenomena are connected at the level of tissue biology. The factors that restrain trophoblast invasion become candidate brakes for metastasis. And the failure mode runs the other way too: when trophoblast invasion is too shallow, the spiral arteries are not properly remodeled, and the result is preeclampsia, the same condition that later marks a woman’s higher cardiovascular risk and may be linked to accelerated biological aging. One temporary organ, studied properly, could have valuable insights when it comes to cancer, cardiovascular disease, and the biology of aging all at once. And yet, for most of medical history it was treated as a disposable afterbirth.

V. The reproductive tract: an ecosystem that forecasts a birth

The vaginal microbiome may contain one of the best early warning signals for preterm birth, a condition that has been difficult to predict before symptoms appear.

Researchers have found that women whose vaginal microbiome is dominated by Lactobacillus crispatus are more likely to carry pregnancies to term. By contrast, lower levels of Lactobacillus and higher levels of bacteria such as Gardnerella and Prevotella are associated with a greater risk of preterm delivery. These patterns were identified in a large pregnancy cohort and supported by additional studies using both microbiome and metabolic data.

This matters because complications from preterm birth are the leading cause of death in children under five worldwide and account for roughly a third of neonatal deaths. Many of those deaths could be prevented if high-risk pregnancies were identified earlier. A routine, low-cost biological sample may contain information that current clinical practice often misses.

The finding also reveals how closely connected different parts of female biology are. Estrogen increases glycogen in the vaginal environment, glycogen supports protective Lactobacillus species, and the metabolites produced by those bacteria help reduce inflammation. Hormones, microbes, and immune signaling are not separate systems. They work together as a single biological network. One reason these discoveries took so long is that medicine often studied each piece in isolation rather than seeing the whole system.

That raises the uncomfortable question: If so many insights were missed in bodies medicine could observe directly, what else is missing in the data systems we now treat as ground truth?

The problem is not only absence. It is compression. Female biology was often captured too thinly: sex as a checkbox, pregnancy as a temporary exception, menopause as a life stage rather than a systemic transition, symptoms as isolated complaints rather than patterns across time. The variables that might have made the biology legible were either not collected, not structured, not linked, or not treated as meaningful enough to change the model.

That is where this becomes an AI governance question.

AI systems do not learn from reality itself. They learn from the reality institutions have recorded. If the record missed the signal, the model inherits the miss. If the dataset collapsed sex, hormones, chromosomes, pregnancy, lactation, menopause, immune state, microbiome, adverse drug reactions, and symptom chronology into a generic patient average, then the model may perform well on paper while failing the populations whose biology was never properly represented.

The danger is not only bias in the familiar sense. It is lost discovery. A model trained on incomplete evidence does not just make worse predictions. It narrows what the system knows how to ask. It may miss the warning signal, flatten the subgroup, misread the risk, misprice the body, or recommend care built around a patient who never existed.

This is the question behind Fit for Whom? Sex-Stratified Data and the Integrity of High-Risk AI.

Join Us In Geneva

In July, Geneva will briefly become one of the places where the future of AI governance is being negotiated.

During the same week, WSIS Forum, AI for Good, and the inaugural UN Global Dialogue on AI Governance will bring governments, standards bodies, companies, researchers, regulators, and civil society into the same city.

That matters because defaults get written in rooms like this. And one default has to change: no high-risk AI system should be certified on aggregate accuracy alone.

Our WSIS session, Fit for Whom? Sex-Stratified Data and the Integrity of High-Risk AI, asks the question every governance process should ask before deployment.

Fit for the average patient?

Fit for the well-coded patient?

Or fit for the people whose data was thin, scattered, delayed, and dismissed?

Sex-stratified data is not a women’s issue. It is an integrity requirement. Disaggregated reporting is how we find failure in the system before it becomes infrastructure. It is how we know whether a model is still attached to reality.

We would love for you to join us in Geneva on July 7:
https://www.itu.int/net4/wsis/forum/2026/Agenda/Session/184

Alongside the WSIS panel, I’ll also be hosting Invisible Value, a small evening gathering in Geneva on the forms of knowledge that rarely appear in formal datasets but quietly shape how people make decisions, seek care, build trust, and navigate everyday life.

If you’ll be in Geneva that week and would be interested in being included, reach me at oriana@femtechnology.org.