Fertility clinics already work with a lot of data. Patients undergoing evaluation or treatment come in for CD3 bloodwork, serial ultrasounds, LH monitoring, and endometrial assessments. What they do not typically have is a continuous picture of what a patient's cycle was doing in the weeks between visits. Clair Health's approach is to fill that gap, not to replace any existing touchpoint.
This post is about what that looks like in practice: what the data actually provides, where it is clinically useful, and where it is not.
The scheduling problem fertility clinics already have
Cycle-based scheduling is one of the persistent operational friction points in fertility care. Most interventions in an IVF or IUI cycle need to be timed relative to cycle day, which requires the clinic to know when a patient's cycle starts. The standard approach is patient-reported period onset with a CD2 or CD3 baseline visit. For patients with regular cycles, this works adequately. For patients with irregular cycles, it creates scheduling chaos.
A patient with PCOS-pattern irregular cycles might go 45 days between periods, and neither she nor the clinic knows when the next cycle will start until it does. Missing the CD2 or CD3 window means waiting for the next cycle. In a context where patients are often paying out of pocket and every delayed cycle represents both financial and emotional cost, the inability to anticipate cycle onset several days in advance is a real problem.
Passive continuous cycle phase monitoring from a wrist biosensor does not solve the PCOS scheduling problem completely. PCOS is one of the populations where wrist-based inference is hardest, and we want to be upfront about that. But for patients with moderately irregular cycles, where cycles vary by 7-10 days in length rather than being wholly unpredictable, continuous wearable monitoring can give the clinic 3-5 days of advance notice that a new cycle is likely approaching based on late-luteal signal patterns. That is enough time to adjust scheduling before the patient calls on day 1.
What the wearable data adds between visits
The more general value proposition for fertility clinics is not cycle-start prediction; it is the inter-visit picture of cycle behavior. Consider a patient who comes in at CD3 for a baseline ultrasound, has a normal-looking baseline, and then returns two weeks later for a day-12 ultrasound. Those two data points do not tell the clinician much about what happened in between. Did the patient have a clear follicular rise and a detectable periovulatory window? Did the pattern look like a normal follicular development trajectory, or was it suppressed or disrupted by something? The wrist data fills this picture with a day-by-day phase inference that connects the two visit endpoints.
This is particularly useful in natural cycle monitoring scenarios and in the evaluation phase before deciding on a treatment protocol. A patient who presents with irregular cycles and reports "I can never tell when I ovulate" is providing useful subjective information, but two or three months of continuous wrist data that shows her actual phase patterns is substantially more useful for protocol planning. Has she been ovulating but with a short luteal phase? Is she showing anovulatory patterns? Are the signals consistent with a PCOS-pattern follicular arrest? This is not a replacement for a serum progesterone or an antral follicle count, but it provides a between-visit behavioral picture that lab tests at isolated time points cannot.
How the data surfaces to clinicians
In the current early access integration, the wrist data appears as a patient data view in the Clair Health portal. For each enrolled patient, the clinician sees a 30-day rolling phase timeline showing daily phase state probabilities, confidence intervals, and signal quality flags for days where data coverage was low or wear time was insufficient.
We made a deliberate choice to show confidence intervals and signal quality flags rather than hiding them to present a cleaner interface. A clinician looking at a cycle timeline that shows a confident luteal phase followed by three days flagged as "low signal quality" is seeing accurate information. Presenting those three days as confidently luteal would be more misleading. The flag prompts a conversation with the patient about whether she stopped wearing the device during that period, which is itself clinically relevant information.
The data view does not generate recommendations, alerts, or treatment suggestions. We are not in the clinical decision support business at this stage. The data is surfaced as additional context that a clinician can interpret using their existing clinical knowledge, not as a system that tries to do the clinical interpretation for them.
What this is not
We want to be direct about the limitations because overstating the clinical value of wrist biosensor data in a fertility context is a real risk in this field.
Wrist phase inference is not equivalent to serum hormone measurement. A phase probability of 0.78 luteal does not confirm that progesterone levels are adequate, that ovulation occurred, or that the luteal phase is clinically normal. It indicates that the patient's biosensor signal pattern is consistent with what the model has learned to associate with luteal biology. That is a meaningful data point; it is not a lab result.
Wrist data is not diagnostic. It does not diagnose anovulation, luteal phase defect, or PCOS. It surfaces a pattern; the clinical interpretation of that pattern requires a clinician with access to the patient's full history, labs, and imaging.
Data aligned with HIPAA-friendly handling practices is a requirement for clinical use, not a differentiator. We have designed the data flow to avoid transferring any personally identifying health record data through the Clair Health system; only the wearable sensor stream is transmitted. But clinics should conduct their own due diligence on data agreements, particularly for any patient population that includes minor patients or patients enrolled in research protocols with specific data governance requirements.
The scenario where it works best today
Based on our early access feedback so far, the scenario where the wrist data is most clearly useful is a patient in a natural cycle evaluation or early investigation phase who has regular or mildly irregular cycles and is motivated enough to wear the device consistently. In this scenario, the 30-day phase timeline gives the evaluating clinician a continuous picture of cycle behavior that substantially contextualizes whatever they find on the CD3 baseline and mid-cycle scan. It helps them move faster from "let's see what her cycle looks like" to "here is what her cycle pattern has been for the past three months."
The scenario where it is least useful is a patient already deep in a stimulated IVF protocol, where the exogenous hormones override the endogenous signal the wrist device is trying to read. During stimulation, the wrist data does not tell you much because what we are measuring is the body's own hormonal rhythm, and stimulation protocols largely suppress that rhythm. We tell clinics this explicitly rather than suggesting they enroll all patients regardless of protocol stage.
What we are building toward is a product that is useful enough across enough of the fertility care journey that it becomes a natural part of how clinics set up monitoring for patients. We are not there yet. But the inter-visit gap in continuous cycle data is a real problem with a data-grounded solution, and we are working through what that solution needs to look like in clinical practice.