The 3-Level Problem: On the gap between health tracking and health understanding
This guest post explores how modern wearables collect so much data, yet fail to deliver true health understanding.
Hello! Popping in here to note that this is a special guest post written in collaboration with two fellow scientists: Nicholas W. Gentry, PhD and Sehresh Saleem, PhD. Their full bios can be found at the end of the article. It is a longer-form, deeper dive than my usual posts, but the message about data accountability in women's health couldn't be more critical. Enjoy the read!
Patients today arrive at clinical appointments with months of high-resolution biometric data. Wearable-derived sleep staging, heart rate variability, skin temperature, and respiratory rate; direct-to-consumer hormone panels and metabolic markers; continuous glucose readings; genetic risk scores. Their physicians have no framework to interpret it, or often ignore it. The data exists, the rigorous evidence to make it clinically useful does not.
Women’s reproductive health is where this gap is most acute. Wearables now track cycle phase alongside skin temperature; direct-to-consumer labs offer AMH, FSH, and estradiol panels without a prescription; apps stitch the data into scores and personalized recommendations. The promise, sold across femtech, is that more information about your body means more clarity, more control, and better decisions.
That promise has driven an extraordinary expansion in personal health data collection. It has not delivered the clarity it advertises.
Is more data the answer?
Nearly half of American adults now own a wearable health tracking device, and that number grows yearly.1 The wearables market is projected to reach $152 billion by 2029, more than double its 2024 size.2 Direct-to-consumer lab testing has proliferated alongside it, with companies offering hormone panels, microbiome assessments, metabolic markers, and genetic risk scores, all accessible without a prescription and all accompanied by the same disclaimer: this does not constitute medical advice.3
Anti-Müllerian hormone (AMH) testing is a useful case study into why this expansion hasn’t translated into clarity. AMH is a ‘functional’ biomarker4 of ovarian reserve, useful in some contexts as a proxy. It is also, in consumer contexts, routinely marketed as a predictor of fertility, a measure of reproductive age, and an indicator of when to freeze eggs, claims that significantly outpace what the evidence supports. AMH correlates poorly with natural conception rates in the general population and, in fact, an ACOG Committee Opinion specifies that serum AMH is not recommended for counseling non-infertile women on their reproductive status.5 Women are paying for tests that deliver a number, a percentile ranking, and, frequently, a low-grade panic without the clinical context that would make the number interpretable.
AMH is just one symptom. Wearables now generate continuous data on heart rate variability, skin temperature, sleep staging, respiratory rate, and “cycle insights.” DTC platforms offer cortisol, thyroid panels, and comprehensive metabolic markers. Emerging women’s health companies are promising to merge the two with continuous hormone monitoring devices. The promise is consistent: more information means better understanding of your health.
Patient generated health data has enormous promise to improve clinical outcomes, but this is only possible when it is integrated into clinical workflows for conditions with established biometrics, validated reference ranges, and reimbursement pathways; contexts like hypertension management, heart failure monitoring, and arrhythmia detection.6 Motivation to share one’s longitudinal health data is simply the first step, and is meaningless if the rest of the infrastructure to receive, process, analyze, and integrate it is missing.
Who is accountable?
The clinical integration problem is downstream of an evidence accountability problem. Physicians cannot incorporate what hasn’t been rigorously validated without assuming liability. This is how the medical system has been designed to work, and for good reason.
For over sixty years, pharmaceutical companies have been required to prove efficacy through controlled trials before bringing a product to market. Over the decades since, that requirement has been reinforced with trial pre-registration, mandatory reporting of results regardless of outcome, and external peer review.7, 8 Each of these was added in response to a specific failure mode: fabricated efficacy, suppressed negative results, undisclosed conflicts. The system is imperfect, but the accountability infrastructure it creates is what makes clinical trust possible. Doctors can prescribe a drug because there is a validated public evidence basis they can evaluate. Trials are pre-specified, ensuring negative results are released and effect sizes are evident. Without these, medicine cannot advance with confidence. The regulatory requirement to publish null results is therefore not a constraint on innovation, but the critical mechanism by which new medical technologies become usable.
Consumer health companies operate under no equivalent accountability structure.9 Datasets are proprietary. Null results disappear into internal servers while positive findings get press releases. If a wearable or biomarker company discovers that one of its measures underperforms, its incentive structure points toward concealment; to protect its reputation, but also to avoid signaling a scientific dead end to competitors. If a pharmaceutical company behaved this way, the response would be regulatory action and justified outrage. In consumer health this has become standard practice.
To hold this industry accountable, you first need a framework for what accountability would actually require at each stage of the chain.
The missing blueprint
The value derived from a health sensor has three levels. Level 1 is sensor accuracy: does the hardware reliably capture a physiological signal? Level 2 is data interpretation: does that signal correspond to something biologically or clinically meaningful? Level 3 is outcomes: does using this data actually improve health? (Figure 1)
The consumer health industry has invested heavily in proving Level 1 and is marketing as though Level 3 is established. The gap between those two claims is where the consumer is left standing, alone, with data and no infrastructure to interpret it (Figure 2).
Step tracking data provides the most illustrative example for the outcome question. Step counting and activity tracking represent the most mature consumer health biometrics, those with the largest database, the longest research history, and the most widespread use. More information, ‘gamification’ of the readout; these are the rationales consumer health companies tout to claim they are improving public health. However, a meta-analysis of dozens of randomized controlled trials showed a different reality. Despite modest short-term increases in physical activity, tracking steps provides no statistically significant improvements in blood pressure, cholesterol, or quality of life, and no clinically meaningful improvements in glucose regulation.10, 11
Not every company in this space is failing in the same way. Apple has moved through many of these levels for some features and earned clinical credibility as a result.12 Companies like Oura and WHOOP have invested in Level 1 and, in some cases, Level 2: their sensors work, and selected signals have been studied for biological correspondence.13 But the Level 3 validation stops where the business model no longer requires it, and the marketing does not stop with it.12 WHOOP marketed its blood pressure feature as delivering “medical grade health and performance insights”, despite an FDA warning letter. They disputed the oversight and declined to remove the feature.14
New consumer health companies focused on hormone testing and biomarker panels occupy even shakier ground. Function Health offers over a hundred biomarkers per panel, including lipids, hormones, inflammatory markers, metabolic indicators, accompanied by color-coded dashboards and percentile rankings. Modern Fertility has been a major driving force behind the spread and overpromised value of AMH testing in fertility and menopause. Supplement stacks, microbiome testing, continuous monitoring capabilities, and others are likewise being heavily marketed directly to consumers today. These companies provide the curious individual the ability to obtain enormous amounts of information on their biology without any evidence that much of this information correlates to improved health outcomes. Rather, the plethora of information now likely adds to the diagnostic confusion that brought most customers to the platform in the first place.15 They have not demonstrated that the panels they sell correspond to actionable clinical meaning for the populations buying them, yet market as if this were self-evident.16 They are selling Level 1 as though Level 3 were implied. Those are not the same thing; genuine biomarkers are stripped of the clinical context required to interpret them, sold with a percentile ranking that implies precision not supported by rigorous science. The customer receives a number, the number generates a feeling, the feeling drives a decision, yet at no point in that chain has clinical interpretation occurred.
More health data is not the answer, it is only the beginning of the sensemaking process. Companies that work through all three levels of validation for their products gain something competitors cannot lobby away: clinical trust. A physician can act on a validated data stream and elicit health improvements where the data alone fails.17 A physician cannot act on a wellness score generated by a proprietary algorithm with no published validation methodology.
What can you do?
The current healthcare system was never built for the possibility that health information would be democratized. Consumers built this market with their purchasing decisions, in response to real shortcomings disserving their needs. However, this also gives them more leverage to shift things than most realize. What does an informed consumer need to do to incentivize the advancement of real science that can positively impact their health?
Primarily, skepticism and awareness. One option is to apply the framework above. Before paying for a consumer health product, ask which of the three levels of validation it has actually demonstrated (Figure 1). Most have invested in Level 1 and stopped there. Some have published Level 2 work. Few have demonstrated Level 3. If unsure of what the data shows, ask friends or look for the opinions of trusted experts in the space. Most times, the flashier a claim, the less likely it is to be grounded in evidence.
When a doctor seems dismissive of wearable or DTC-gathered data, the most likely explanation is not indifference. It is that they are operating within a system that has given them no reimbursement pathway for interpreting longitudinal datasets, no liability framework for acting on consumer-grade data, and no training standard for evaluating what these platforms actually measure.
A market shaped by more selective consumers could compel changes in the system we have now. More builders should be expected to register studies prospectively and submit validation methodology to external research and review; the longitudinal datasets consumer health companies have accumulated are a genuine scientific resource, and treating those findings as trade secrets holds back the entire field.18 Companies would then ideally compete on validated outcomes, rather than biomarker count and dashboard polish. Clinicians, for their part, would reap the benefits of improved liability protections that would let them engage with the data their patients are already bringing in, and open the door to better training in this area. If there is more appetite for it in the market, regulators could feel greater pressure to build more applicable accountability frameworks for consumer health tools; despite popular conception, the FDA for example has shown recent willingness to innovate (Figure 3).19
None of this happens by accident. The market is shaped by what consumers reward and what they refuse to pay for. Right now, it is being rewarded for selling Level 1 as though it were Level 3.
The gap can be closed
A permanent shift in how health data is generated and distributed is underway. The consumer health industry has demonstrated that people will pay for information about their own bodies. It has not yet demonstrated that it is willing to be accountable for what that information means.
The issue is not that every company must replicate Apple’s regulatory pathway. It is that every company collecting health data has a responsibility to be honest about where on this chain their evidence actually stops, and to keep pushing further. The gap between what has been validated and what is being sold is not a marketing problem. It is an accountability problem. Selective analysis and reporting on internal data is not accountable science.
The stack of results is still growing. The question is whether anyone will be accountable for what it means.
About the guest authors:
Nicholas Gentry, PhD, is CEO and Cofounder of LifeAhead, a startup commercializing evidence-based forecasts of reproductive lifespan. Their mission is to build LifeAhead with rigor and accountability as a core design constraint, not a market afterthought.
Sehresh Saleem, PhD, is the founder of Ovealth, a women’s health company advancing ovarian tissue preservation as a pathway to extend reproductive and hormonal lifespan. A geneticist and biotech leader by training, she is focused on translating rigorous science into accessible options for women seeking greater agency over fertility, menopause, and long term health.
If you want to see this framework applied to real-world clinical cases, check out Dr. Lara Zibners’ fantastic recent piece: "Fine, I'll Admit It: Sometimes Watching Hormones in Real Time Would Actually Help." She breaks down the line we are drawing here, explaining why real-time data tracking is a game-changer for IVF and complex medical conditions, but total nonsense when packaged as a passive wellness score for healthy women.
Lots of love from me, yourIVFbaby!
I’m a reproductive scientist and scientific communications consultant helping women’s health companies translate complex science into credible communication. If you’re building in this space, I’d love to connect → jordan@jordanmachlinphd.com
Thanks for reading! This post is public so feel free to share it.
If you want notifications on new posts please subscribe - it’s free!
References:
1. Nagappan, A., Knowles, M. & Krasniansky, A. Put a ring on it: understanding consumers’ year-over-year wearable adoption patterns. Rock Health.
https://rockhealth.com/insights/put-a-ring-on-it-understanding-consumers-yearover-year-wearable-adoption-patterns/ (2024).
2. MarketsandMarkets. Smart wearables market trends, size & growth outlook 2030. MarketsandMarkets. https://www.marketsandmarkets.com/ResearchInsight/smartwearables-market.asp (2026).
3. Kalokairinou, L., Choi, R., Wei, N. & Wexler, A. Policies of US companies o ering direct-to-consumer laboratory tests. JAMA Intern. Med. 183, 1275–1278 (2023).
4. Moolhuijsen, L. M. E. & Visser, J. A. Anti-Müllerian hormone and ovarian reserve: update on assessing ovarian function. J. Clin. Endocrinol. Metab. 105, 3361–3373 (2020).
5. American College of Obstetricians and Gynecologists’ Committee on Gynecologic Practice. The use of antimüllerian hormone in women not seeking fertility care. ACOG Committee Opinion No. 773. Obstet. Gynecol. 133, e274–e278 (2019).
6. Treadwell, J. R., Reston, J. T., Rouse, B. et al. Automated-Entry Patient-Generated
Health Data for Chronic Conditions: The Evidence on Health Outcomes Technical Brief No. 38 (Agency for Healthcare Research and Quality, Rockville, MD, 2021).
7. Greene, J. A. & Podolsky, S. H. Reform, regulation, and pharmaceuticals — the Kefauver–Harris amendments at 50. N. Engl. J. Med. 367, 1481–1483 (2012).
8. Brazil, R. Illuminating ‘the ugly side of science’: fresh incentives for reporting negative results. Nature. https://doi.org/10.1038/d41586-024-01389-7 (2024).
9. Coravos, A., Doerr, M., Goldsack, J. et al. Modernizing and designing evaluation frameworks for connected sensor technologies in medicine. npj Digit. Med. 3, 37 (2020).
10. Ferguson, T., Olds, T., Curtis, R. et al. E ectiveness of wearable activity trackers to increase physical activity and improve health: a systematic review of systematic reviews and meta-analyses. Lancet Digit. Health 4, e615–e626 (2022).
11. Hodkinson A, Kontopantelis E, Adeniji C, van Marwijk H, McMillian B, Bower P, Panagioti M. Interventions Using Wearable Physical Activity Trackers Among Adults With Cardiometabolic Conditions: A Systematic Review and Meta-analysis. JAMA Netw Open. 2021 Jul 1;4(7):e2116382.
12. CB Insights. Apple is going after the healthcare industry, starting with personal health data. CB Insights. https://www.cbinsights.com/research/apple-healthcarestrategy-apps/ (2019).
13. Tang, H. Is the Oura ring considered a medical device? Complete FDA classification guide. Vertu. https://vertu.com/lifestyle/is-the-oura-ring-considered-a-medicaldevice-complete-fda-classification-guide/ (2025).
14. Hagen, J. WHOOP pushes back on FDA over blood pressure feature.
MobiHealthNews. https://www.mobihealthnews.com/news/whoop-pushes-backfda-over-blood-pressure-feature (2025).
15. Svensson, T., Madhawa, K., Hoang, N. T., Chung, U.-I. & Svensson, A. K. Validity and reliability of the Oura Ring Generation 3 (Gen3) with Oura sleep staging algorithm 2.0 (OSSA 2.0) when compared to multi-night ambulatory polysomnography: a validation study of 96 participants and 421,045 epochs. Sleep Med. 115, 251–263 (2024).
16. The Lancet. Direct-to-consumer medical testing: an industry built on fear. Lancet 404, 991 (2024).
17. Shih, P., Ding, P., Carter, S. M. et al. Direct-to-consumer tests advertised online in Australia and their implications for medical overuse: systematic online review and a typology of clinical utility. BMJ Open 13, e074205 (2023).
18. Hicks JL, Altho T, Sosic R, Kuber P, Bostjancic B, King AC, Leskovec J, Delp SL. Best practices for analyzing large-scale health data from wearables and smartphone apps. npj Digit Med. 2019 Jun 3;2:45. doi:10.1038/s41746-019-0121-1. PMID:
31304391; PMCID: PMC6550202.
19. Kroetsch, A. FDA’s real-time clinical trials pilot could transform drug development. The Clinical Trials Abundance Blog. https://www.clinicaltrialsabundance.blog/p/fdas-real-time-clinical-trials-pilot (2026).
Yes! Love the diagram that shows so clearly the 3 levels we are operating within and yet we still need the "so what" whether that is "what to do about it" or "what changes for patients because of it."