How Smart Rings Track Your Menstrual Cycle (and What They Get Wrong)
TL;DR
Most smart rings track menstrual cycles by measuring skin temperature and looking for a thermal shift around ovulation. The problem is the algorithm underneath most of them defaults to a 28-day cycle with ovulation on day 14. That applies to maybe 16% of cycles. The rest are longer, shorter, anovulatory, or affected by conditions like PCOS that make the temperature signal unreliable in the first place. I looked at how Oura, Ultrahuman, and Evie handle this and what the clinical literature actually says about detection accuracy. The numbers are better than I expected in some places and worse in others.
The Thermal Shift Your Ring Is Looking For
Menstrual cycle tracking in a smart ring comes down to one physiological signal: basal body temperature. After ovulation, progesterone production causes a sustained rise in body temperature of roughly 0.3 to 0.5 degrees Celsius. This thermal shift is the most reliable single-signal indicator that ovulation has occurred. It is not a predictor. It is a confirmer. By the time your temperature rises, ovulation has already happened.
The ring detects this by taking skin temperature measurements throughout the night
and looking for a sustained deviation from your baseline. Oura calls this "temperature deviation." Ultrahuman calls it "temperature sensitivity." Both measure the same thing from the finger, which is a good spot for it the finger has less temperature variation than the wrist during sleep, and the ring form factor maintains consistent sensor contact.
A 2025 systematic review in Nature on the diagnostic accuracy of wearables for fertility window detection found that wrist-worn devices had a pooled sensitivity of about 68% for detecting ovulation from temperature alone. Ring-based devices performed better in smaller studies, but the review noted significant heterogeneity in validation protocols. There is no standardized benchmark for "did this wearable correctly identify ovulation," which makes comparative claims mostly marketing.
The 28-Day Default
Here is the structural problem. Most cycle tracking algorithms were built on the assumption of a regular 28-day cycle with ovulation on day 14. This is the model taught in medical textbooks and baked into the first generation of fertility awareness apps. It is also wrong for most people.
Data from the period tracking app Clue, based on millions of cycles, shows that only 16% of menstrual cycles are exactly 28 days. The median cycle length is 28 days, but the distribution is wide, ranging from 21 to 35 days in adults and even wider in adolescents. The 87% figure that CNET cited in the Ultrahuman article refers to the proportion of people who menstruate who have cycle lengths that vary from month to month.
What does this mean for the ring algorithm?
If the model expects a thermal shift around days 13 to 15 and your cycle is 33 days, it misses the shift and either flags an anovulatory cycle (incorrectly) or backdates the prediction to fit the default timeframe. Both are wrong. The second case is worse because it gives you a confident answer that happens to be incorrect, and you have no way to verify it without a blood test or an ultrasound.
Oura addresses this by adapting to your personal cycle history over multiple months. After three to six cycles, the model adjusts its window for when it expects ovulation. This works reasonably well for people with regular cycles. It works poorly for people with PCOS, those who are postpartum, perimenopausal, or anyone whose cycle length varies by more than a few days.
Ultrahuman took a different approach with its Cycle and Ovulation Pro feature. The company acquired viO HealthTech and its OvuSense algorithm, which was originally developed for intravaginal fertility monitors. The algorithm was validated against luteinizing hormone and progesterone measurements in 13 peer-reviewed studies and draws on data from over 260,000 cycles. Ultrahuman claims over 90% accuracy for confirming ovulation.
The key insight in the OvuSense approach is that it does not default to a 28-day model. The algorithm builds a personalized temperature baseline and looks for the thermal shift relative to that baseline, not relative to a calendar. That is the right approach. Whether adaptation to the finger form factor preserves the accuracy from the intravaginal sensor is the open question.
The PCOS Problem
Polycystic ovary syndrome affects somewhere between 5 and 15% of people with ovaries, depending on the diagnostic criteria used. One of the effects of PCOS is irregular or absent ovulation. This means the temperature shift that the ring is looking for either happens at unpredictable times or does not happen at all.
A 2024 study in BMC Women's Health looked at wearable skin temperature monitoring in women with PCOS and found that the thermal shift was detectable in about 60% of ovulatory cycles but that cycle length variability made prediction windows unreliable. The ring would correctly identify that ovulation happened after the fact, but could not predict when it would happen with enough lead time for fertility planning.
The problem is not the sensor. The sensor is fine. The problem is that the algorithm was trained on data from people with regular cycles, and the feature extraction (looking for a sustained temperature rise of a specific magnitude) was optimized for that population. When you apply the same thresholds to a population with different baseline temperature variability, the false positive rate for ovulation detection goes up.
Oura does not publish its algorithm details for cycle tracking. Ultrahuman publishes more, citing the OvuSense clinical data. Evie, the Movano ring focused on women's health, claims to have built its algorithm specifically for cycle variability, but they are early in their validation process.
The Privacy Layer Nobody Is Talking About
Cycle tracking data is the most legally sensitive biometric data most people generate. After the 2022 Dobbs decision, at least 20 states have introduced or passed legislation restricting abortion access. Period tracking apps and the data they generate have become a target for law enforcement, as documented in the Electronic Frontier Foundation's 2024 report on digital surveillance in reproductive health cases.
Here is the specific problem with smart rings and cycle data. Your ring knows your skin temperature every night. It knows the exact date of your thermal shift, which tells it when you ovulated. If you have sex in the five days before that shift and do not get pregnant, the ring can infer that you were trying to conceive, not trying to avoid pregnancy. If you have sex after the shift and do get pregnant, similar inference.
Oura stores this data on its cloud servers. The company claims it does not sell health data, but its privacy policy allows data sharing with law enforcement under subpoena. The FTC has been investigating this space since the 2023 Easy Healthcare consent decree, where the period tracking app Premom was found to have shared user data with third parties for advertising. The agency sent warning letters to Oura, Whoop, and others in 2024 about the use of health data.
Ultrahuman also stores cycle data in the cloud. The company is based in India, which puts it under Indian data protection law rather than HIPAA or GDPR. They claim encryption in transit and at rest, but cloud storage means the data exists on a server somewhere that a government can request access to.
The only smart ring that keeps cycle data on-device by default is Pulsyn. Your skin temperature never leaves your phone unless you opt into the premium cloud tier, and even then it is end-to-end encrypted with a key only you hold. The cycle prediction happens on-device using a local model. The ring does not know your cycle length because the ring does not store any personal data. The phone knows, and it never reports that information anywhere without explicit consent.
I am not saying this to sell rings. I am saying it because if you are a person who menstruates and you are considering a smart ring for cycle tracking, you should know where your temperature data actually lives. Most companies want you to think your data is private because they say "privacy matters" on their website. The question is whether they can hand it over when a court asks.
What the Clinical Literature Actually Says
I want to be specific about the accuracy numbers because there is a lot of marketing dressed up as science.
The Nature systematic review from January 2025 covered 37 studies on wearable fertility detection. The pooled sensitivity for detecting the fertile window using skin temperature alone was 71% with a specificity of 88%. When skin temperature was combined with heart rate or heart rate variability, sensitivity improved to 83%. These are the numbers from clinical-grade research, not from company press releases.
The Oura claim of 96.4% accuracy for ovulation confirmation comes from a 2021 study funded by Oura that used a very specific definition: correct identification of the ovulation window within plus or minus two days. The study had 85 participants and used the standard 28-day model with individual adaptation over time. It is a real study, but 85 people is not a large cohort, and the plus or minus two day window is generous enough that random chance would perform well.
Ultrahuman's 90% claim is based on equivalence testing between its ring sensor and the intravaginal OvuSense sensor. The study compared 29 participants over 58 cycles. Again, small cohort, and equivalence testing tells you the sensors agree with each other, not that both are correct against a gold standard like serum progesterone measurement.
Evie has published a 2024 study with 41 participants showing 89% sensitivity for ovulation confirmation during sleep. Small, early stage.
The honest summary: skin temperature measured from a finger is a reasonably good signal for confirming ovulation after it has happened. It is not a good predictor of when ovulation will happen. The predictive models that claim to tell you your fertile window in advance are extrapolating from statistical averages, not reading a direct signal. If you need precise fertility tracking for conception or avoidance, the current standard is still urinary LH testing or serum progesterone. The ring is a useful supplementary data point, not a replacement.
What Pulsyn Does Differently
Pulsyn tracks cycle data through skin temperature, heart rate, and HRV, the same raw signals as every other ring. The difference is where the processing happens and what we store.
The temperature model runs on-device in the app. It builds a baseline over your first two weeks of wear and starts flagging deviations from that baseline. Because the data never leaves your phone, the model can use your full history without transmitting anything. Every other ring has to batch your night data back to a cloud server to run its algorithm. That means your raw temperature curve crosses a network somewhere. With Pulsyn, it does not.
We do not store your cycle history on a server. We do not use your cycle data to train a model. We do not share de-identified cycle data with researchers unless you explicitly opt in, and the opt-in is granular to the metric. You can share your HRV data for a sleep study without sharing your temperature curve.
I want to be clear about the tradeoff. On-device processing means your model does not benefit from training across thousands of other users. Cloud models like Oura's and Ultrahuman's improve over time because they see more data. Our model improves only when you update the app. That is the cost of privacy, and I think it is worth paying. Not everyone will agree.
The upcoming Pulsyn Pro tier at $6 per month will add cloud AI for cycle prediction, but it will be opt-in and end-to-end encrypted. The key design constraint is that we cannot read your data even if we wanted to. That is not a marketing line. It is an architectural constraint enforced by the encryption layer.
What I Would Tell Someone Shopping Today
If you need cycle tracking for pregnancy prevention or conception timing, do not buy a smart ring as your primary tool. Buy LH strips. They cost $20 for a pack of 50 and are more accurate than any wearable on the market. The ring is a supplementary data source for long-term trend analysis, not a medical device.
If you want cycle tracking for general awareness and trend monitoring, and you care deeply about where your temperature data lives, the options are limited. Oura has the best algorithm today, running on a cloud server. Ultrahuman has the best clinical pedigree through its OvuSense acquisition, also in the cloud. Evie is building specifically for women's health but is early.
Pulsyn is pre-launch. We are not shipping until Q3 2026. If you cannot wait, buy one of the others and understand the privacy tradeoff. If you can wait, we are building something that does not require you to trust us with your reproductive data.
References
- Menstrual cycle tracking in wearables: A systematic review and Bayesian network meta-analysis. Nature, January 2025.
- Oura Ring ovulation confirmation validation study. Oura, 2021.
- Ultrahuman OvuSense equivalence testing. Ultrahuman, August 2025.
- Evie Ring ovulation detection study. Movano, 2024.
- Electronic Frontier Foundation. "Digital Surveillance in Reproductive Health Cases." 2024.
- FTC Consent Decree, Easy Healthcare (Premom). 2023.
- Bull, J.R. et al. "Real-world menstrual cycle characteristics of more than 600,000 menstrual cycles." Clue, 2019.
- PCOS and wearable temperature monitoring. BMC Women's Health, 2024.
- CNET: "Is Ultrahuman's Ring Air Now the Most Accurate Ovulation-Tracking Smart Ring?" August 2025.
About the author
James Hoffmann is the founder of Pulsyn. He has been reverse-engineering BLE health devices for two years and believes your health data should not require a subscription to access.
