In short: anna is an evidence-based perimenopause app. It reads the data your wearable already records, matches the patterns against a library of over 1,000 clinician-reviewed rules drawn from more than 300 research papers, and gives you one explained tip each morning, with the study behind it. Think of it as a menopause app built for the perimenopause years: pattern recognition over your own data, reviewed by a clinician, not AI and not a symptom diary.
This page is the long version: how the rules were built, what data anna reads, how a signal becomes a sentence you can act on, and, just as importantly, what anna does not do. If you point one person at a single page to understand what anna is, this is the one.
How the rules were built
anna runs on over 1,000 rules, drawn from more than 300 peer-reviewed papers and reviewed by a clinician before any of them reach you. A rule is a documented link between something measurable in your data and something meaningful in perimenopause, together with the guidance that follows from it.
Each rule started as a question in the research. What actually happens to sleep across the transition? To resting heart rate? To body temperature at night? We read the evidence for each one, extracted the finding, and wrote it into a rule only where the research supported it. An advising clinician then reviewed the rules for accuracy and safety. Where the evidence is thin or mixed, the rule says so, or it is left out. That editorial discipline, cite it or cut it, is the whole point.
This is deliberately unglamorous work, and it is what separates a tool built on evidence from one built on assertion. Most apps in this space tell you what they think. anna shows you why, and where the why comes from.
What data anna reads
anna reads passive signals that your Apple Watch or compatible wearable already records, through Apple Health, with your permission. Nothing here asks you to log a mood score at midnight or fill in a diary. The signals include your sleep, your resting heart rate, your heart rate variability and your skin temperature, among others.
These are not arbitrary choices. They are the measures the research has tied most directly to the hormonal shifts of perimenopause. Resting heart rate and heart rate variability move measurably with hormonal state across the transition (de Zambotti et al., 2017). Skin temperature and the physiology behind hot flushes are objectively detectable by body sensors rather than by self-report alone (Hunter and Mann, 2010). anna works with the signals your watch is genuinely good at capturing, and leaves the rest alone.
The translation logic
A raw number is not an insight. Your resting heart rate rising by a few beats over a fortnight means nothing on its own, and everything in the right context. anna's job is the translation in between.
It works like this. anna watches for a pattern across your signals over time, not a single reading. It matches that pattern against the rule library. When a rule applies, it surfaces one explained heads-up for the day, in plain language, with the reasoning attached: here is what shifted, here is what the research links it to, here is one thing you might do with that. One tip, not a wall of dashboards. The aim is a single useful thing you did not know when you woke up, rather than more data to interpret yourself.
A quick example. Say your resting heart rate has crept up by a few beats a night across a fortnight, while your sleep has quietly become more broken. On their own, neither number means much. Read together, they match a known pattern in the perimenopause research, and anna surfaces that as your tip for the day, in plain language, with the study it rests on, and one small thing you might try.
The citation system
Every tip anna gives is anchored to its source. When anna tells you something, you can see the research it rests on, named and dated, not a vague appeal to studies show. This is the feature we are least willing to compromise on.
It matters for two reasons. It lets you check anna's reasoning rather than take it on trust, which is how a health tool should treat an adult. And it means anna can only say what the evidence will support, because a claim with no citation does not ship. The citation system is not a footnote. It is the constraint that keeps the rest of the product honest.
What anna does not do
Being clear about the edges is part of being trustworthy.
- anna does not diagnose, treat or prescribe. It detects patterns and suggests. It is a tool for self-awareness, not a medical device, and it is not a substitute for your doctor.
- anna is not a symptom diary. The point is that your wearable is already recording the signals, so you do not have to become your own data-entry clerk.
- anna does not predict your final period or forecast your future. At launch it describes what your signals are doing now and what that tends to mean, and it does not pretend to a certainty the science does not have.
- anna is not marketed as artificial intelligence. It uses pattern recognition over your wearable data, reviewed by a clinician. That framing is honest about what is happening under the bonnet, and we would rather be accurate than fashionable.
The research domains
anna's library is organised into more than twenty clinical and physiological domains. Each rule sits inside one of them, and each domain rests on its own body of evidence. A representative source from several of the main areas:
Sleep. Disturbed sleep and insomnia affect 40 to 60% of women during the menopausal transition, and women who develop menopausal insomnia show objectively measured sleep loss, not just the perception of it (Baker et al., 2015).
Cardiovascular signals. In the early menopausal transition, resting heart rate rises and heart-rate-variability measures fall, and both track the hormonal state, the same signals a consumer wearable reads (de Zambotti et al., 2017).
Vasomotor symptoms. Hot flushes are driven by oestrogen withdrawal narrowing the body's thermoneutral zone, so that even small rises in core temperature trigger flushing and sweating, a physiological event, not a feeling (Casper, UpToDate, 2026).
Cognition. Oestrogen receptors are distributed throughout the brain and concentrated in regions tied to memory, which is the biological basis for oestrogen's role in cognition during the transition (Grodstein, UpToDate, 2026).
Weight and metabolism. Abdominal visceral fat increases through perimenopause and levels off around the final period, marking a real shift toward central fat distribution rather than a failure of willpower (Wildman and Sowers, 2011).
Reproductive hormones. Oestrogen is protective of heart health before menopause, and follicle-stimulating hormone rises as ovarian function declines, serving as a marker of the transition itself (El Khoudary, 2017).
The full library spans these areas and more, from bone health to mood to the genitourinary changes women are rarely warned about. The symptoms guide is where a lot of it becomes practical.