Perimenopause Testing Meets AI: What Changes Next

Perimenopause affects over 1 billion women globally, yet most care still runs on a familiar (and frankly exhausting) pattern: describe symptoms, try a treatment, wait, adjust, repeat. That trial-and-error loop burns months or years—right when people are juggling careers, teens, aging parents, and the usual end‑of‑year stress that December tends to amplify.

A new entrant in women’s health, Willbe, is betting that perimenopause shouldn’t be treated as a vague life stage you “get through,” but as a biological transition with measurable signals. Their newly announced test, FemGene, positions perimenopause as an epigenetic reprogramming event—a shift in how genes are expressed as hormones change—and claims it can predict how someone will metabolize and respond to bioidentical hormone therapy (BHRT).

For readers following our AI in Healthcare and Medical Technology series, this is the part that matters: once care becomes measurable, it becomes modelable. Genetics, hormones, symptoms, wearables, and outcomes can be turned into AI-assisted decision support that reduces guesswork—if it’s built responsibly.

Perimenopause isn’t “just hormones”—it’s a systems shift

Perimenopause is commonly framed as falling estrogen and progesterone. That’s true, but incomplete. The better framing is: hormones are the messaging system, and perimenopause changes the message frequency, timing, and strength. Your brain, sleep, metabolism, skin, bones, and cardiovascular system all receive different “instructions.”

Calling it an epigenetic reprogramming event is a strong statement, but it captures something clinicians see every day: two people with similar ages and similar lab values can have wildly different symptoms and treatment responses.

What “epigenetic reprogramming” means in practice

You don’t need a molecular biology degree for this to be useful. Here’s the practical translation:

• Genes don’t change, but which genes are turned “up” or “down” can shift with hormonal changes.
• Those shifts can influence inflammation, sleep regulation, insulin sensitivity, mood, skin changes, and recovery.
• Treatment response varies because the body’s “settings” vary—especially around hormone metabolism and receptor sensitivity.

This is exactly the kind of complexity where medicine tends to default to “try and see.” It’s also exactly the kind of complexity where better data + careful AI can outperform intuition alone.

What FemGene is trying to do (and why it’s worth watching)

FemGene is presented as a test designed to “decode how women uniquely experience perimenopause and respond to BHRT,” using Willbe’s proprietary approaches branded as Hormogenomics and Hormogenetics.

Their promise is straightforward: predict treatment response earlier, so fewer people bounce between doses, formulations, or therapies while symptoms keep disrupting life.

The “Hormonal Archetypes” idea: helpful, with a caveat

Willbe also promotes a framework called Hormonal Archetypes, essentially grouping individuals into categories rooted in genetics to guide care pathways.

I’m generally in favor of structured clinical frameworks—they help clinicians and patients communicate and make decisions faster. The caveat is important though: archetypes are only as good as the evidence behind them.

Here’s what good looks like:

• Clear explanation of which variants are used and why
• Validation showing the archetypes predict real outcomes (symptom reduction, adherence, side effects)
• Performance measured across age ranges, ethnicities, comorbidities, and medication histories

If FemGene can show that level of transparency and validation, it becomes more than a marketing concept—it becomes a serious clinical tool.

Where this fits in real care pathways

Perimenopause care typically includes some combination of:

• Symptom tracking (sleep, hot flashes, mood, bleeding patterns)
• Hormone testing (often inconsistent and time-sensitive)
• Lifestyle interventions (nutrition, resistance training, alcohol reduction)
• BHRT or other pharmacologic options where appropriate

A genetics-informed test doesn’t replace clinical judgment. It front-loads useful constraints: “These options are more likely to work, these are more likely to cause issues, and here’s how fast we should titrate.” That’s valuable if it reduces time-to-relief.

Where AI can actually help (and where it can’t)

AI in women’s health gets oversold when it promises a single magical prediction. The more realistic win is AI as a coordinator—connecting lots of small signals into a coherent, personalized plan.

AI-assisted diagnostics: pattern recognition across noisy data

Perimenopause is notoriously hard to “diagnose” with a single test because:

• Symptoms overlap with thyroid dysfunction, anemia, depression/anxiety, sleep apnea, and burnout
• Hormone levels fluctuate day-to-day
• People present differently depending on stress, weight changes, and medications

A strong AI-assisted approach would combine:

• Genetic markers (like those used in tests such as FemGene)
• Longitudinal symptom reports (weekly trends beat one-off questionnaires)
• Wearable data (sleep fragmentation, resting heart rate, temperature trends)
• Labs where available (lipids, HbA1c, ferritin, thyroid markers)

Done right, AI doesn’t “diagnose perimenopause” in isolation. It flags probability and patterns, helps rule out look-alikes, and supports clinician decision-making.

Targeted treatment: matching therapy to metabolism and side-effect risk

The biggest opportunity is reducing trial and error with BHRT by predicting:

• Metabolism differences that affect hormone levels and symptom control
• Higher likelihood of side effects that lead to stopping therapy
• Which dosing strategies may require tighter follow-up

This is where genetics can be a meaningful input. AI can take genetic signals, mix them with real-world follow-up outcomes, and continuously improve recommendations.

Where AI can’t help: missing context and poor-quality inputs

AI fails fast when:

• Data is sparse (one-time snapshots)
• Self-reported symptoms are inconsistent because tools aren’t user-friendly
• Models aren’t validated across diverse populations
• The system doesn’t integrate clinician judgment (especially around contraindications)

A practical rule: if a platform can’t explain why it’s recommending something in plain language, you shouldn’t trust it.

The business case: women’s health is underserved—and that’s changing

Women’s health has been underfunded and under-researched for decades. Perimenopause care is a perfect example: massive population impact, relatively little precision.

From a healthcare system perspective—especially in Ireland and the UK—this matters because unmanaged perimenopause symptoms contribute to:

• Reduced work productivity and increased absenteeism
• Higher use of primary care visits for non-specific symptoms
• Increased risk profiles over time (cardiometabolic and bone health pathways)

If precision approaches improve adherence and reduce symptom duration, they don’t just improve quality of life. They reduce churn in the system.

What “precision longevity medicine” should mean (not just a slogan)

Willbe frames perimenopause as “the front door to female longevity.” That’s a compelling line, and it can be true if it translates into prevention that’s measurable:

• Earlier intervention for bone density decline (strength training + clinical monitoring)
• Earlier identification of cardiometabolic drift (lipids, blood pressure, insulin resistance)
• Better sleep stabilization (a major driver of mood and metabolic health)

Longevity talk gets fluffy quickly. The version I trust is the version tied to measurable outcomes and follow-up.

Practical guidance: how to evaluate perimenopause tests and AI tools

If you’re a clinician, digital health leader, employer benefits manager, or simply someone navigating perimenopause, here’s what I’d look for before trusting a new test or platform.

Questions to ask any perimenopause genetic test provider

1. What does the test actually measure? Specific variants, pathways, and intended clinical interpretation.
2. What outcomes has it predicted in validation? Symptom improvement, time-to-stable dosing, side effects, discontinuation.
3. Who was included in the evidence? Age ranges, ethnic diversity, comorbidities, medication use.
4. How are results explained? Clear, actionable guidance beats vague “risk” scores.
5. How is data stored and used? Opt-in consent, deletion options, and no quiet resale of sensitive data.

Questions to ask any AI-driven women’s health platform

• Is it a decision support tool or is it pretending to replace clinical care?
• Does it integrate with real workflows (telemedicine, GP, specialist referral)?
• How often does it prompt follow-up when symptoms change?
• Does it handle safety checks (contraindications, red flags, abnormal bleeding pathways)?

A simple “better care” checklist for perimenopause

These aren’t glamorous, but they work:

• Track symptoms weekly for 8–12 weeks (sleep, mood, bleeding, hot flashes)
• Get baseline labs that support differential diagnosis (iron status, thyroid markers, metabolic markers)
• Treat sleep as a primary outcome, not an afterthought
• If using BHRT, plan structured follow-up (2–6 week adjustments are common)

Tools like FemGene may help you start in a better place, but ongoing follow-up is where outcomes are won.

What this signals for AI in healthcare—and why Ireland should pay attention

FemGene is one product, but the broader trend is bigger: women’s health is becoming data-rich. When that happens, AI has a legitimate role—especially in:

• Remote monitoring and telemedicine triage
• Personalized treatment selection and titration support
• Early detection of risk patterns tied to aging-related conditions

Ireland has strong fundamentals for this wave: medtech expertise, research capacity, and an innovation ecosystem that can build clinically responsible tools. The risk is that we import platforms without demanding validation in local care pathways.

Perimenopause care improves fastest when biology, behavior, and follow-up are treated as one system—not separate checkboxes.

FemGene’s positioning—perimenopause as a measurable biological transition—pushes the conversation in a useful direction. The next step is proving it with outcomes, not slogans.

If you’re building or buying AI-assisted diagnostics for women’s health, set the bar high: transparent evidence, real-world validation, and workflows clinicians will actually use. That’s how we get to care that feels less like guesswork and more like medicine.