Over the past weeks, a lot of smart people, including Amy Beckley PhD and Susanne Mitschke, have already pointed out that Clair Health’s wristband does not measure hormones. They are right, but I want to go further: explain what these devices actually do, and point to the bigger news from Germany almost no one is talking about. First, the difference the whole debate turns on. Estrogen, progesterone, LH and FSH circulate at very low concentrations and do not cross intact skin in a way a wrist sensor could read. So a wristband cannot read the molecules themselves. What it can read are downstream effects: skin temp, heart rate, HRV, sleep. The same signals ŌURA and WHOOP already read. Clair then runs an algorithm that estimates the hormones from those proxies. It never touches the hormone. The founders have a sharp defense for that. They redefine the terms. A CGM, they argue, does not measure glucose directly either. In their words, "neither measures the target analyte directly. Both infer it from correlated signals." So hormone inference, they say, is the same thing. Except that it is not. A CGM DOES measure glucose. An enzyme on the sensor reacts with the actual glucose molecules in your interstitial fluid, producing a current that tracks how much is there. Different fluid than blood, yes. The analyte itself is still measured. Clair never meets the hormone. It reads heart rate and temperature, different quantities, and a model fills in the rest. A different fluid is not the same as a different quantity. For me, calling that the world's first continuous hormone monitor is a deception of the women it is built for. We are at the very start of this field. If the loudest early product wears a label it has not earned, women lose trust before the real thing ever ships. One bad definition can set a whole industry back years. Which is why the real news matters. SPRIND - Bundesagentur für Sprunginnovationen, GER's Fed Agency for Breakthrough Innovation, has opened a 40 M € Continuous Hormone Monitoring Challenge, designed and led by Ida Tin, the founder of Clue. And it draws the line this field needed. To qualify, a device has to measure the hormone itself. Continuously, on-site, in a biological matrix like ISF, sweat or saliva. Surrogate signals like HR or temperature are explicitly not accepted as a stand-in for a hormone level. That one rule is the entire debate. It tells every team, including mine, what the bar actually is. I build biochemical sensors for a living. I'm a founder in this field, which is why I know how hard true molecular sensing is. That difficulty is the point. It separates a real measurement from a confident guess, exactly what women's health cannot afford to blur. Innovation here is welcome. So is ambition. But the bar is now written down. Measure the hormone, or do not call it a hormone monitor. Women deserve devices they can trust. And trust starts with describing accurately what our technology actually measures.
Dr. Agnes Musiol that is exactly right! Thanks for writing this. Optimism about tech innovation that sounds great is not enough - we need scientific rigor, scientific humility, scientific proof, before a startup goes out making sweeping claims that precede proof, and before investors direct capital to scientifically unsubstantiated claims. I shared my thoughts on the day in the below article too, which underscores many of the points you make, and extends to more. This is a good case study for the VCs directing capital... https://www.linkedin.com/pulse/continuous-hormone-monitoring-biological-utility-chasing-valcheva-4siif/?trackingId=8Nb7VjEpS%2F63sw9ipXIr%2Bw%3D%3D
Audrey Sherman
1mo
If you’re going to say that you’re monitoring a chemical-ie. A Molecule… you probably better measure for the molecule and not a product or an effect of it. That’s like saying because the grass is wet- it rained or is raining…what about dew or even a sprinkler. And worse would be to take actions based on that data.
Anton Pchelkin
1mo
Worth adding from years of building sensing into wearables: even honest, direct measurement on the wrist is hard — continuous sweat sensing and bioimpedance both fight a constant battle with signal quality. And that's when you're measuring a real quantity. Inferring a picomolar hormone from weaker downstream proxies is a much harder problem than "continuous hormone monitor" suggests. But the optimist in me wins here: the molecular path is arriving — aptamer sensors, microneedle ISF sampling, microfluidic matrix capture — all aimed at touching the actual molecule, continuously. The wrist was never the ceiling. Direct molecular contact is, and that's now being solved.
Tahir Ahmed
1mo
This is so well explained, Innovations are appreciated but I’m just curious about the audacity most of these wearable startup’s and even health tech platforms have, literally measuring the same biomarkers and data - all they do is interpret/infer it in the fanciest ways possible, raise huge money, trend for quite sometime with that hype and don’t really deliver real outcome. Huge claims and titles feel good but like you mentioned, it’s not even about the debate, its about the standard you set, the real outcome you deliver and lastly it’s about what people can trust.
What surprises me most is how often companies can raise impressive amounts of money while making claims that seem, at best, scientifically fragile. At ENDOless, we took the opposite path: everything we build is grounded in published research, validated questionnaires, and real-world patient data logic. And yet, proving that rigor sometimes feels harder than selling a shiny promise. I honestly wonder sometimes: is the problem the science, the market, or the founder trying to explain it? Because women’s health does not need more confident guesses dressed as innovation. It needs trust, evidence, and the courage to say clearly what a technology can — and cannot — measure.
Aora
1mo
This distinction matters beyond hormones. “Measuring the target” and “estimating a state from correlated proxies” can both be useful, but they should not be marketed as the same thing. We think about this a lot while building Aora, a behind-the-ear neurotech wearable for continuous cognitive state monitoring: what is directly sensed, what is inferred, and what decision the user can responsibly make from it. i.e. latest Apple Watch will tell you "your VO2 Max is 46" but it should say that's not your VO2 Max but rather a inference, with proven significant error. This aims to gain the masses but risks losing customer's trust in the long run
The more interesting question isn't how the hormone level is obtained, it's how accurate and reliable the estimate is. Many measurements in engineering and medicine are inferred rather than directly measured. If a variable can be estimated from other signals with sufficient accuracy, precision and robustness across different people and conditions, then it can still provide substantial clinical value.
Vandana Pramod
1mo
Thank you for explaining this so clearly. I think this distinction will become increasingly important as AI becomes more integrated into healthcare. AI can help connect patterns, but the quality of those insights will always depend on being honest about the underlying data.
Dr. Agnes Musiol Thanks for sharing this reflection. It fits tightly with the work I’m doing on next‑generation female‑physiology systems: https://phase-awaremodelingolimpialopezfaba.my.canva.site/ The real problem isn’t just technical... it’s behavioural. When a device claims biochemical access it doesn’t have, it mislabels technology and it distorts the psychological contract with women (who often are actively seeking something trustworthy to anchor their decisions). Curious where you think the pressure should be applied first... terminology discipline, scientific thresholds, or user‑psychology impact?
Thank you Dr. Agnes Musiol, as the PPG technology partners of TI AFE and ADI VSM, working on PPG for the last 20 years, I was originally amazed at how we never came across monitoring Estrogen, progesterone, LH and FSH with PPG. As Audrey Sherman summed up nicely, you can't say that because the grass is wet, that it definitely had rained.