Wearable Technology for Menstrual Cycle Tracking

The body has been quietly broadcasting cycle data for decades. We just didn't have the hardware to listen continuously, or the algorithms to make sense of what we heard. Wearable cycle tracking changes that, but the technology is uneven, the marketing is frequently ahead of the evidence, and the privacy stakes are higher than most users realize. Understanding what these devices actually measure, how well they measure it, and where they fall short is the only way to choose the right tool for a specific goal.

How the four-phase cycle maps onto the biosignal patterns wearables detect

The menstrual cycle produces measurable, repeatable changes across several body systems: skin temperature, resting heart rate, heart rate variability (HRV), respiratory rate, and sleep architecture. These are passive signals. The body generates them continuously; wearables capture them at scale and over time, then run machine learning algorithms trained on large datasets to weight each biomarker differently across cycle phases.

A large-scale analysis published in npj Digital Medicine, covering more than 11,500 participants across nearly 46,000 cycles, quantified what that looks like in practice. Resting heart rate peaks around cycle day 26 and bottoms out around day 5. HRV follows the inverse pattern. Skin temperature shifts approximately 0.5°C post-ovulation. HRV drops roughly 12% and resting heart rate rises approximately 8 bpm from the follicular to luteal phase.

Those numbers sound precise, and they are, at the population level. At the individual level, the picture is more nuanced.

Menstruation and the follicular phase

Days one through five represent the physiological baseline: lowest resting heart rate, highest HRV, temperature at its nadir. This is the reference point from which deviations are measured throughout the rest of the cycle. The follicular phase that follows, roughly days six through thirteen, sees a gradual estrogen rise, but the biosignals remain relatively stable and low. That stability is actually the problem: the follicular phase is the hardest for wearables to characterize, because there is not much happening in the signals to distinguish it from a neutral biological state.

The ovulation window

This is the most clinically consequential moment in the cycle, and it is where the technology's asymmetry becomes most apparent. A temperature rise of approximately 0.5°C confirms ovulation has occurred. HRV begins to drop. But that confirmation is retrospective: the temperature rises after the egg has already been released. Predicting ovulation 24 to 48 hours in advance, which is what matters for timed conception or confident avoidance, is the harder problem. Most wrist-worn devices solve for confirmation, not prediction.

The luteal phase

Days fifteen through twenty-eight, progesterone dominant, are where wearables perform most reliably. Sustained temperature elevation, elevated resting heart rate, suppressed HRV, changes in sleep quality and respiratory rate: the luteal phase is the most physiologically distinctive window in the cycle, and the one that biosignal monitoring was essentially built to detect. Respiratory rate and sleep architecture function as supporting signals here, adding dimensionality when temperature or heart rate data is noisy on a given night.

The asymmetry between confirmation and prediction is not a flaw in any particular device; it is a structural feature of the underlying physiology. Anyone using cycle data for contraception versus conception planning should understand this distinction before choosing a device.

The accuracy picture: what the clinical evidence actually shows

A January 2026 systematic review and Bayesian network meta-analysis published in npj Digital Medicine, covering 27 studies across 8 databases, found that wearable digital technology achieved a pooled accuracy of 0.88, with a 95% confidence interval of 0.86 to 0.90, a sensitivity of 0.79, and a specificity of 0.80.

That 0.88 pooled accuracy sounds strong. Interrogate it further, and the sensitivity figure becomes the more important number. Sensitivity of 0.79 means roughly one in five fertile days goes undetected. Depending on whether the goal is achieving or avoiding pregnancy, that gap ranges from a disappointment to a meaningful risk. The figure is also a pooled average across devices and methodologies, which means high-performing devices are averaging with lower-performing ones.

A separate systematic review from early 2024 found that most devices performed well for detecting the luteal phase, fertile window, and menstruation, but that performance varied significantly by device and study methodology. Device-level claims are more granular: one wrist-worn ring's algorithm has been reported to correctly identify ovulatory cycles with accuracy in the mid-90s per a 2025 study; a dedicated fertility-focused sensor claimed fertile window identification in the high 90s and ovulation date accuracy above 93% of cycles per company-cited peer-reviewed data.

One important distinction the aggregate figures obscure: most published accuracy numbers are retrospective. They measure how well a device identified that ovulation occurred after the fact, using the full cycle's data. Prospective accuracy, predicting the fertile window one to two days before it opens, is lower across every device category and rarely the headline figure in marketing materials.

Natural Cycles offers a useful benchmark precisely because it has been through regulatory scrutiny the others have not. Its FDA-cleared labeling reports 98% effectiveness with perfect use and 93% with typical use, figures that place it in the same range as combined hormonal contraception. That comparison is only meaningful because of what the clearance process required: prospective efficacy data, post-market surveillance, and ongoing reporting obligations.

How form factor shapes what a device can and cannot track

Hardware placement determines signal quality, and signal quality determines what an algorithm can do. This is not a minor technical detail; it is the reason two devices with similar marketing can produce meaningfully different results.

Smart rings

Rings position sensors against finger vasculature, where blood vessels run close to the surface and are less insulated by subcutaneous fat than the wrist. That proximity produces more consistent temperature and pulse readings overnight. Because most ring users wear them during sleep, and sleep is when physiological signals are most stable and least contaminated by activity and stress, rings tend to generate cleaner data than devices worn only during the day. Temperature accuracy for the leading ring devices is reported in the range of 0.1°C, which is the resolution needed to detect the half-degree ovulatory shift.

Smartwatches

Watches are worn during the day and overnight, capturing a broader behavioral and physiological picture: activity, stress responses, and sleep. The tradeoff is wrist placement, which is noisier for temperature readings than the finger. The wrist has more adipose tissue between the sensor and the blood vessels, and it is more exposed to ambient temperature fluctuation. Leading watch manufacturers use skin temperature sensors alongside heart rate data to build retrospective ovulation estimates. At least one major watch manufacturer secured FDA approval for its fertility tracking feature in 2025, though the company itself positions it as an estimation tool rather than a medical device.

Arm and overnight bands

The bracelet form factor, worn only overnight, positions sensors differently than a ring or watch, tracking multiple physiological parameters simultaneously including pulse rate and physiological stress indicators. For users who do not sleep comfortably in rings or watches, this form factor improves adherence, and adherence is the variable that most directly determines data quality over time.

Intravaginal sensors

This is where the signal fidelity difference becomes most pronounced. Core body temperature, measured internally, is more stable than peripheral skin temperature. It is unaffected by room temperature, sleep position, or ambient conditions. Sensors in this category take readings at high frequency throughout the night, generating a continuous temperature curve rather than a single overnight average. That resolution is what enables same-cycle predictive capability: detecting the subtle pre-ovulatory temperature pattern in the current cycle rather than using historical averages to project forward. The tradeoff is obvious. Invasiveness versus resolution is a real consideration, and the population willing to use an intravaginal sensor nightly is smaller than the population willing to wear a ring.

The form-factor decision is ultimately an adherence decision. The most accurate sensor in the world produces useless data if a user stops wearing it after three weeks.

What the main devices on the market actually offer in 2025–2026

Oura Ring 4

The market-leading smart ring for cycle tracking updated its Cycle Insights algorithm in December 2025, with the company reporting meaningfully improved period prediction accuracy overall and notably better performance for users with irregular cycles. Temperature accuracy sits at 0.13°C. The ring integrates directly with Natural Cycles, which matters: pairing a well-validated sensor with a regulated algorithm is a different product than either component alone.

Ultrahuman Ring AIR

Ultrahuman acquired OvuSense in 2025, folding fifteen years of intravaginal sensor research into a paid feature called Cycle and Ovulation Pro. The company claims ovulation confirmation accuracy above 90% and positions the product specifically for users with PCOS and endometriosis, citing compatibility with users whose cycles fall outside the standard length range. The OvuSense acquisition is significant because it brings same-cycle predictive methodology into a ring ecosystem. Independent validation of accuracy claims in irregular-cycle populations is still limited.

Apple Watch Series 8 and later

The broadest installed base of any device in this category by a substantial margin. Apple's Cycle Tracking app logs symptoms alongside overnight temperature and heart rate data. The ecosystem matters as much as the sensor: integration with Apple Health, compatibility with Natural Cycles, and third-party app support create a health data infrastructure that extends well beyond cycle tracking. FDA approval for the fertility tracking feature was secured in 2025. The watch remains an estimation tool, and Apple's own framing reflects that, but the combination of scale and ecosystem makes it relevant to any discussion of the category.

Ava Bracelet

Designed specifically for fertility awareness rather than general health. Worn overnight, it tracks five parameters simultaneously including physiological stress and resting pulse rate, then syncs in the morning. Its narrower scope is also its argument: every design decision is oriented toward cycle and fertility data rather than general wellness.

Garmin

Garmin's cycle tracking lives in the Menstrual Cycle card within Garmin Connect, covering period history, cycle and fertile window predictions, ovulation test logging, mood and flow tracking, and pregnancy support. In March 2026, Garmin announced Natural Cycles integration for several compatible smartwatch models. This integration extends a regulated algorithm to a new hardware ecosystem, which illustrates how the regulatory pathway for iterative AI updates is designed to function.

OvuSense standalone

For users who prioritize predictive ovulation accuracy above all else, the standalone intravaginal sensor remains the highest-resolution consumer option. Niche, but the data it produces is qualitatively different from what wrist devices can generate. Now also available as the methodology underlying Ultrahuman's premium cycle feature.

Natural Cycles as the regulatory benchmark for what FDA clearance means in practice

Natural Cycles has the most detailed public regulatory history of any product in this category, which makes it the most useful reference point for evaluating what "FDA-cleared" actually means versus what marketing language implies.

The company received De Novo FDA clearance in 2018, the first digital contraceptive to achieve it. The years since have involved hardware integrations with both a leading smart ring and a leading smartwatch, a Predetermined Change Control Plan approved in 2024, over-the-counter and prescription availability in 2025, and a sixth clearance in 2026 for a next-generation fertility algorithm. The NC° Band, launched under the 2024 FDA pathway, reads overnight temperature passively; users no longer need to take a manual morning reading. The hardware and the cleared algorithm function as a single regulated system.

The PCCP pathway is worth understanding in its own right. It allows Natural Cycles to update its algorithm as evidence accumulates without filing a new 510(k) for each iteration. This is significant because it means the regulatory framework now explicitly accommodates iterative machine learning improvement, which is how these algorithms actually develop. Clearance is not a one-time event; it is a framework for ongoing accountability.

Most other wearables marketed for cycle tracking are wellness devices, not cleared medical devices. That is not necessarily a problem, but it means the accuracy and safety claims have not been subjected to the same evidentiary standard. Natural Cycles is useful not because every user needs a contraceptive-grade device, but because it defines what clinical rigor in this category actually looks like.

Where wearables underperform and which users face the biggest gaps

The 2026 npj Digital Medicine meta-analysis identified menstrual regularity as a key driver of accuracy variation: wearable digital technology showed meaningfully lower detection accuracy in populations with irregular menstruation. This finding matters because irregular cycles are common, especially among the populations most likely to be actively tracking their fertility.

Why irregular cycles are harder

Algorithms trained predominantly on cycles in the 24 to 35-day range have less training signal for cycles outside that window. The physiological patterns are still present in irregular cycles; they are shifted or compressed in ways that standard prediction models handle less gracefully. The fertile window is harder to locate when baseline cycle length is unpredictable, and temperature shifts may be subtler or occur at unexpected points in the cycle.

The 2024 systematic review offers an important nuance here: even for users with irregular cycles, wearables likely outperform purely date-based diary methods. The physiological signals exist regardless of calendar regularity. The gap is relative to regular-cycle users, not relative to tracking nothing.

PCOS and endometriosis

Anovulatory cycles, cycles in which no ovulation occurs, produce no temperature shift to detect. A wearable may not reliably distinguish an anovulatory cycle from a late ovulation, which creates real ambiguity for users managing PCOS. Ultrahuman's OvuSense integration is explicitly positioned to address this population, but independent accuracy data in this group remains sparse.

The confirmation versus prediction gap

Most devices confirm ovulation after it has occurred. The temperature rise is a retrospective signal by definition. Users trying to time conception need advance notice of the fertile window, not confirmation that it has closed. Only intravaginal sensors with high-frequency overnight readings currently claim same-cycle predictive capability, and that claim is most established for the intravaginal sensor category.

Sensor performance and skin tone

Photoplethysmography sensors, which most wearables use to measure heart rate and oxygen saturation, perform differently across skin tones. This is an acknowledged limitation in the broader wearable literature. Its specific impact on cycle tracking accuracy has not been well characterized in published research, but it is a known variable.

Adherence and onboarding

Algorithms need several cycles of data to personalize predictions. New users should not expect full accuracy from day one. Data quality degrades with missed nights. The devices that are easiest to wear consistently, not necessarily the most sophisticated sensors, tend to produce the most reliable long-term data for individual users.

Data privacy and what happens to cycle data after it leaves the device

Menstrual and fertility data is among the most sensitive health information a consumer generates. It can reveal pregnancy status, pregnancy loss, contraceptive use, and sexual activity. That sensitivity has legal implications that have become concrete since the Supreme Court's 2022 decision in Dobbs v. Jackson Women's Health Organization.

In states where abortion access is restricted, cycle and pregnancy data has been identified as potential legal evidence. Law enforcement requests to health app companies have occurred. This is not a theoretical concern; it is documented. The question of what happens to cycle data after it leaves a device is therefore not a secondary consideration. For many users, it is the primary one.

What to evaluate

The key distinctions are practical. Is the data stored locally or in the cloud? Is it used for advertising or sold to third parties? Does the company's privacy policy explicitly address law enforcement requests and, if so, what is the company's stated posture? Is the data covered by HIPAA? Most consumer health apps are not HIPAA-covered entities, which means the regulatory protections many users assume apply do not.

Natural Cycles, as a regulated medical device, operates under different data handling obligations than a general wellness app. But regulatory status is not equivalent to data protection. Readers should verify current privacy policies directly rather than inferring data protection from clearance status.

The practical guidance is specific: look for end-to-end encryption, a clear data deletion option, and an explicit stated policy on government data requests. Some applications offer anonymous or pseudonymous modes, which reduce the identifiability of stored data.

One critical distinction: device choice and app choice are separable decisions. A well-validated sensor paired with a poorly governed application creates a risk that sensor accuracy cannot address. Evaluating the data infrastructure is as important as evaluating the hardware.

Matching the right device to the reader's actual health goal

Three distinct use cases carry three distinct accuracy requirements. Listing them in ascending order of stakes clarifies what level of rigor each situation actually demands.

General cycle awareness

Almost any wearable with overnight temperature and heart rate tracking is sufficient. A smartwatch already owned, or a fitness ring used primarily for sleep and recovery, will surface the broad cycle patterns: the luteal phase shift in temperature and heart rate, the follicular phase baseline, the rough period window. Adding symptom logging alongside biosignals creates a richer picture without requiring new hardware. The accuracy stakes here are low; the goal is pattern awareness, not precise timing.

Conception support

Accurate fertile window prediction matters more, and the device choice should reflect that. A smart ring paired with a regulated algorithm represents a well-evidenced option for users with regular cycles. For users with irregular cycles, PCOS, or endometriosis, a device with explicit same-cycle predictive capability, specifically the intravaginal sensor methodology whether standalone or integrated into a ring platform, is the better fit. Expect a calibration period of several cycles before predictions stabilize.

Hormone-free contraception

Only FDA-cleared methods belong here. Natural Cycles, whether used as a standalone app, as the NC° Band system, or integrated with compatible hardware, is the only wearable-adjacent option that has cleared the regulatory bar comparable to other contraceptive methods. Its 93% typical-use effectiveness means user behavior still matters substantially. The gap between perfect use and typical use is real, and it is larger than the gap for methods that remove user action from the equation entirely.

Irregular cycles, PCOS, endometriosis

Prioritize devices with explicit irregular-cycle validation and set honest expectations. Even the best available technology performs less accurately in this population per the 2026 meta-analysis. That does not mean tracking is without value; physiological signals are present regardless of cycle regularity. It means that algorithmic outputs should be interpreted with more caution, not less engagement.

The factors beyond accuracy

Form factor preference and sleep habits determine adherence, and adherence determines data quality. Subscription costs for premium cycle features are real and recurring. Ecosystem fit, whether data flows into Apple Health, Google Health, or Garmin Connect, affects how useful the data is across an entire health picture. Data privacy posture is a device and app selection criterion, not an afterthought.

One final point: no device replaces clinical care for diagnosing cycle disorders. Wearable data can be a genuinely useful record to bring to a clinician. Patterns flagged by an algorithm are a starting point for a conversation, not a diagnosis. The technology is better than it has ever been. It is still not medicine.