How Wearables Are Changing Personal Health Tracking: 7 Things the Data Doesn’t Tell You

By Dr. Narayan Rout | Author | Researcher |    Next Human Series  ·  30 min read  ·  Published: June 27, 2026

Contents In This Research Pillar

Table of Contents

1. Introduction
2. 1. How Big Has Wearable Health Tracking Actually Gotten?
3. 2. Can a Smartwatch Actually Measure Your Blood Sugar?
4. 3. Why Does the Sensor Inside Your Smartwatch Have a 32-Year-Old Accuracy Problem?
5. 4. Is This a ‘Wicked Problem,’ or Can It Just Be Fixed in the Next Software Update?
6. 5. Can Tracking Your Own Health Data Actually Make You Less Healthy?
7. 6. So Which Wearable Metrics Should You Actually Trust?
8. 7. What’s Actually Coming Next — and Why Does It Raise a New Kind of Trust Question?
9. The Quest Sage Insight
10. What You Can Do With This
11. Conclusion: A Real Tool, With Real and Specific Limits
12. Frequently Asked Questions: Wearable Health Tracking
13. References and Sources
14. Further Reading on Related Topic

In February 2024, the FDA issued a warning that should have been a much bigger story than it was: no smartwatch or smart ring on the market is authorized to measure blood glucose. Not approximately. Not with a caveat. None. This matters because, at the exact same time that warning was being issued, a growing number of consumer devices were being marketed with language that strongly implied exactly this capability — and millions of people were already making real decisions, about real meals and real medication timing, based on numbers their wrist was quietly generating.

This is the actual story of wearable health technology in 2026, and it’s considerably more interesting than either the breathless marketing version or the dismissive “it’s all snake oil” version. The global wearable market hit $86.78 billion in 2025 and is genuinely, measurably changing how people relate to their own bodies — generating real clinical benefit in documented cases, while simultaneously carrying a specific, named, 32-year-old measurement bias that most users have never heard of, and producing a real, clinically recognized form of anxiety in people who track too much, too closely, without the context a clinician would normally provide. This article works through what these devices actually get right, what they get specifically and predictably wrong, and why “is my smartwatch accurate” is the wrong question to be asking in the first place.

The scale here is worth stating precisely, because it changes how seriously the accuracy questions in the rest of this article deserve to be taken. This is no longer a niche fitness-enthusiast product category.

The global wearable technology market was valued at $86.78 billion in 2025, is projected to grow to $96.44 billion in 2026, and is projected to reach $231.43 billion by 2034 — a compound annual growth rate of 11.60%. This isn’t simply more units sold; it represents a genuine shift in how healthcare itself operates. A 2025 systematic review published in JMIR found wearable devices in clinical care now provide patient-centered continuous monitoring, enhanced real-time activity feedback, faster treatment adjustment based on continuous data streams, and meaningfully reduced unnecessary clinic visits through remote monitoring. A separate scoping review of 179 clinical research studies, covering a combined 10.8 million participants, found wearables are now integrated into 71.5% of observational medical research — meaning this technology has moved from consumer gadget to genuine clinical research infrastructure within roughly a decade. (Ref. 1)

This scale is precisely why the accuracy questions examined in the rest of this article matter considerably more than they would for a niche product. A measurement bias affecting a small specialty device is a footnote. The same bias, built into a sensor technology now worn by a meaningful fraction of a global population making real health decisions from it daily, is a genuinely significant public health question.

No — and this is worth stating with the same directness the FDA itself used, because the gap between marketing implication and regulatory reality here is genuinely significant.

In February 2024, the FDA issued a safety notice stating plainly that it “has not authorized, cleared, or approved any smartwatch or smart ring that is intended to measure or estimate blood glucose values on its own,” specifically warning people living with diabetes against relying on such devices for actual medical decisions. This warning exists precisely because the consumer market had, by that point, produced a real wave of devices marketed with language strongly implying exactly this capability, despite none meeting the clinical accuracy threshold genuine glucose monitoring requires. (Ref. 2) The contrast with what does work is instructive: FDA-cleared continuous glucose monitors like the Dexcom G7 use a small, minimally invasive sensor filament — not an optical or purely external sensor — and achieve genuine, clinically validated accuracy, with the G7 specifically offering up to 15.5 days of continuous wear and direct smartwatch integration for display purposes only.

The precise lesson here, worth carrying into every other section of this article: the accuracy gap isn’t about “wearables” as one undifferentiated category. It’s about the specific measurement method underneath a specific claim. A wearable showing you glucose-adjacent data from an FDA-cleared CGM sensor is a genuinely different, and genuinely trustworthy, claim than a smartwatch implying it can estimate glucose through its wrist-mounted optical sensor alone.

This is the single most important, and most under-discussed, fact in this entire article, and it deserves to be stated with the same precision its original researchers used.

Martin J. Tobin and Amal Jubran, publishing in the European Respiratory Journal in June 2022, titled their paper with the finding itself, stated directly: “Inaccuracy of pulse oximetry in darker-skinned patients is unchanged across 32 years.” Pulse oximetry — the technology estimating blood oxygen saturation by shining light through skin and measuring how much is absorbed — has been documented to systematically overestimate oxygen levels in patients with darker skin tones since the 1980s, an issue that gained global public attention during the COVID-19 pandemic, when accurate home oxygen readings became a matter of real clinical urgency. (Ref. 3) A 2023 cohort study of 24,504 hospitalized COVID-19 patients found this overestimation pattern was associated with Black patients being significantly less likely to receive needed treatment compared to white patients with equivalent actual underlying oxygen levels — an adjusted odds ratio of 1.65, meaning a real, measurable clinical consequence, not merely a theoretical measurement quirk.

Here is the connection most casual coverage of wearables misses entirely: consumer smartwatches and fitness trackers use this exact same optical pulse-oximetry principle for their SpO2 and, in several models, their heart-rate sensors. A documented, three-decade-old, repeatedly-confirmed measurement bias is not a hospital-only problem — it is built into the identical sensor technology now worn daily by a meaningful fraction of a global population, generating the SpO2 and heart-rate numbers many wearable users treat as straightforwardly objective. The table below summarizes the documented timeline.

Year	Development	Significance
1980s	First clinical documentation of pulse oximetry skin-tone bias	Issue identified, but received limited mainstream attention
2020-2022	COVID-19 pandemic brings global attention; NEJM and ERJ publish confirming research	Tobin and Jubran’s 32-year ‘unchanged’ finding published
Dec 2024	JAMA publishes ‘wicked problem’ analysis	Industry-wide, $2B-market-implicating call for coordinated action
Jan 2025	FDA releases updated draft guidance for manufacturers	Regulatory response begins; public comment period through March 2025

It’s worth answering this precisely, because the framing matters for what a reader should actually expect going forward.

Shachar, Drabo, Iwashyna, and Ferryman, writing in JAMA in December 2024, used the specific term “wicked problem” to describe addressing racial and ethnic bias in pulse oximeters — a deliberate academic term for a problem that resists simple, one-time technical fixes because it’s embedded in manufacturing standards, regulatory testing protocols, and a $2 billion existing device industry simultaneously. (Ref. 4) The FDA’s response, a January 2025 draft guidance update for manufacturers on gathering more representative clinical data across diverse skin pigmentations, is real, dated regulatory motion — but draft guidance with a public comment period closing in March 2025 is, honestly, an early-stage regulatory step, not a solved problem with an already-implemented fix.

One important clarification deserves equal weight here, because precision matters for how this finding should actually be understood: a 2022 commentary in the New England Journal of Medicine noted directly that “the term ‘racial bias’ always refers to decisions that are influenced by a person’s race. Medical devices such as pulse oximeters are blind to color and cannot exhibit such a bias” in the intentional sense. The disparity is a documented, measurable performance gap rooted in how the optical sensor technology interacts with melanin concentration in skin — not evidence that any specific company or device was built with discriminatory intent. Both things are true simultaneously: the harm is real and measurable, and it emerged from a genuine technical and testing-protocol gap rather than deliberate design.

This is the section that surprises most readers, and it’s worth taking seriously precisely because it inverts the entire premise wearable marketing is built on.

A January 2026 Bloomberg investigation into health anxiety stemming from wearable data identified a genuine, growing clinical phenomenon: users generating, by conservative estimates, over 500 unique health data points per day across two or more devices — heart rate variability readings every five minutes, overnight blood oxygen measurements, skin temperature trends, glucose response curves after every meal — increasingly making significant lifestyle and even medical decisions based on metrics they lack the clinical training to correctly interpret. (Ref. 5) This connects to a specifically named, recognized phenomenon in sleep medicine: “orthosomnia,” in which excessive concern over sleep-tracker data measurably worsens a person’s actual sleep quality and anxiety — meaning the act of tracking a metric can become a more significant problem than whatever the underlying metric originally showed.

The core issue here isn’t that the data is wrong in every case — several wearable metrics, examined honestly in the next section, are genuinely reliable. The issue is that consumer wearables typically provide a number without the clinical context a trained professional would normally supply alongside an equivalent reading: what’s the normal range for someone your age and fitness level, how much does this metric naturally vary day to day, and what confidence interval should you place around a single reading. Stripped of that context, an entirely normal, ordinary fluctuation can read, to an anxious or undertrained eye, like an alarming signal.

The honest answer is genuinely specific, not a single yes-or-no verdict — and this is precisely the right question to replace “is my wearable accurate” with.

The fourth-generation Oura Ring, released in late 2025, is documented to reliably measure sleep stages, heart rate variability, resting heart rate, body temperature trends, blood oxygen, and menstrual cycle tracking. It is, with equal documentation, explicitly inaccurate for continuous heart rate during intense exercise above 150 BPM on its ring form factor, and it does not measure blood pressure, blood glucose, or provide ECG functionality at all — not an oversight, simply outside what this hardware was built to do. A 2024 Mayo Clinic study comparing Apple Watch Series 7 vital-sign readings against in-room anesthesia monitors during 292 actual endoscopic procedures found measurable concordance for certain vitals under controlled clinical observation — real, positive evidence for specific metrics in a specific, monitored context, not a blanket endorsement extending to every use case. (Ref. 6)

The pattern across all of this validation research is consistent: wearable accuracy reliably degrades during high-intensity movement, for users with darker skin tones specifically on pulse-oximetry-dependent metrics (per Section 3), and for any metric, like blood glucose, with no validated non-invasive consumer measurement method at all. Resting heart rate and sleep-stage tracking, by contrast, hold up considerably better across the validation literature. The honest framework: ask which specific metric, under which specific condition, has been independently validated — not whether the device as a whole deserves trust or distrust.

The current frontier of this technology isn’t a more accurate sensor. It’s AI increasingly interpreting existing sensor data into new, higher-stakes claims — and this raises a genuinely new question current regulatory frameworks aren’t yet fully built to answer.

A March 2026 update to Oura’s women’s health AI model added cycle prediction, fertility window estimation, and pregnancy monitoring features, leveraging the ring’s existing temperature-sensing hardware rather than any new physical sensor. (Ref. 7) This represents a genuine and important shift in what “wearable accuracy” even means going forward: the question is no longer only whether a sensor correctly measures a physical quantity like temperature or heart rate, but whether an AI model’s interpretation of that existing data into a new health prediction — a fertility window, an illness-risk score, a recovery-readiness number — has itself been validated to the same rigor the underlying hardware sensor originally was. Current consumer marketing rarely distinguishes these two layers clearly for the user, and current FDA validation frameworks, built historically around hardware performance testing, are still actively catching up to a world where the most consequential claim a device makes may come from a software model layered on top of perfectly accurate raw sensor data, rather than from the sensor itself.

Here is the argument I think this research actually supports, stated as a claim rather than hedged: wearable accuracy is not a property a device either has or lacks. It is a variable that shifts systematically depending on three separate factors examined throughout this article — what specifically is being measured, under what physical conditions, and on whose body. The FDA’s glucose warning, the Tobin-Jubran pulse-oximetry finding, and the Oura Ring’s documented 150-BPM ceiling are not three unrelated facts about three different limitations. They are the same underlying lesson, stated three times: a wearable’s accuracy claim is only ever as good as the specific validation behind that specific claim, for that specific measurement, and treating the device as a single trustworthy or untrustworthy object obscures exactly the distinction that actually matters.

I think the genuinely original synthesis worth drawing from this research is this: the same technology that has measurably improved clinical care — continuous monitoring, earlier warning signals, reduced unnecessary clinic visits, per the JMIR review in Section 1 — is, through the identical underlying sensor mechanism, also the source of a documented, three-decade-old equity gap and a newly documented form of data-driven anxiety. These are not contradictory findings requiring a verdict on whether wearables are “good” or “bad.” They are the predictable result of deploying a genuinely powerful but specifically limited measurement technology at a scale of hundreds of millions of users, faster than the validation, equity correction, and clinical-context layers around that technology have managed to keep pace. The device is not lying to you. It is reporting, with documented and specific limits, a physical measurement — and the actual skill this era requires is learning to read that report with the same care a clinician would, rather than the blind confidence the marketing invites.

• Never use a smartwatch or smart ring’s blood glucose estimate, if a device claims to offer one, for any actual medical decision — per Section 2, the FDA has explicitly stated no such device is currently authorized for this purpose.
• If you have darker skin and rely on a wearable’s SpO2 or blood-oxygen reading for any health-related decision, know about the documented bias in Section 3 and weight that specific reading with appropriate caution, particularly during any acute illness.
• Before trusting any specific wearable metric, ask which exact measurement it is and under what condition — per Section 6, ‘is this accurate’ should always become ‘is this specific metric, under this specific condition, independently validated.’
• If you notice yourself feeling more anxious rather than more informed after checking your wearable data, per Section 5’s orthosomnia finding, that’s a real, recognized pattern — consider reducing checking frequency rather than assuming more vigilance is automatically healthier.
• When a wearable introduces a new AI-driven health prediction feature (fertility windows, illness risk, recovery scores), per Section 7, ask specifically whether that prediction has its own validation evidence, separate from the underlying sensor’s own accuracy — the two are not automatically the same claim.

Wearable health technology has genuinely transformed personal health tracking — a $86.78 billion market in 2025, real documented clinical benefits, and continuous monitoring capability no previous generation had access to. It has also, through the exact same underlying sensor technology, carried forward a documented, 32-year-old measurement bias affecting darker-skinned users, an explicit FDA warning against a specific, widely marketed capability that doesn’t actually exist yet, and a genuinely new form of data-driven anxiety in users who track more than they can clinically interpret.

The governing argument worth carrying forward from this article: accuracy is not a single verdict to render on a device, but a specific question to ask of every individual measurement it produces — what is being measured, under what condition, validated against what standard, on whose body. A wearable that gets this asked and answered correctly, metric by metric, is a genuinely powerful health tool. A wearable trusted uncritically, as a single undifferentiated source of objective truth about your body, is exactly how a real, useful signal turns into either a missed warning or a manufactured worry.

1. Preventive Medicine Daily (2026). Wearable Technology and Health Tracking Statistics 2026. Global market valuation, growth projections, and 2025 JMIR systematic review on clinical wearable integration. preventivemedicinedaily.com

2. GoodRx. Do Blood Sugar Monitor Watches Work? FDA’s February 2024 safety notice on smartwatch and smart ring glucose monitoring. goodrx.com

3. Tobin, M.J. and Jubran, A. (2022). Inaccuracy of pulse oximetry in darker-skinned patients is unchanged across 32 years. European Respiratory Journal. doi.org

4. Shachar, C., Drabo, E.F., Iwashyna, T.J., and Ferryman, K. (2024). Addressing racial and ethnic bias in pulse oximeters — a wicked problem. JAMA. As reported in AJMC, Racial and Ethnic Bias in Pulse Oximetry Is Failing Patients. ajmc.com

5. AI Magicx (2026). AI Health Wearables in 2026: The Complete Guide to Smart Rings, Continuous Monitors, and What the Data Actually Tells You. Citing Bloomberg’s January 2026 investigation into wearable-induced health anxiety. aimagicx.com

6. Accuracy and role of consumer facing wearable technology for continuous monitoring during endoscopic procedures. PMC. Mayo Clinic study comparing Apple Watch Series 7 against in-room anesthesia monitors, 292 procedures. ncbi.nlm.nih.gov

7. FDA (2024-2025). Draft guidance on pulse oximeter performance evaluation across diverse skin pigmentations, released January 6, 2025. As reported in TechTarget. techtarget.com

8. New England Journal of Medicine (2022). More on Racial Bias in Pulse Oximetry Measurement. Clarification on device-level versus intentional bias. nejm.org

9. Fawzy, A. et al. (2023). Cohort study of 24,504 hospitalized COVID-19 patients on pulse oximetry overestimation and treatment disparity by race. As cited in The Cardiology Advisor. thecardiologyadvisor.com

10. Rout, N. Generative AI’s Impact on Humanity. TheQuestSage.com, Sl 64. Companion piece on the broader societal implications of AI-driven personal technology referenced in Section 7. thequestsage.com

11. Rout, N. Why AI Ethics Matters for Everyday Users. TheQuestSage.com, TQS-2026-147. Companion piece on documented AI bias cases, directly relevant to this article’s discussion of validation gaps in AI-driven health predictions. thequestsage.com

12. Rout, N. Sleep Deprivation Epidemic. TheQuestSage.com, Sl 37. Companion piece on sleep health, directly relevant to this article’s discussion of orthosomnia in Section 5. thequestsage.com

P10 — The Next Human

• Why AI Ethics Matters for Everyday Users (TheQuestSage.com, TQS-2026-147) — The companion piece on documented AI bias cases, directly relevant to this article’s discussion of AI-driven health prediction validation gaps.
• Generative AI’s Impact on Humanity (TheQuestSage.com, Sl 64) — A companion piece on the broader societal implications of AI-driven personal technology.
• Sleep Deprivation Epidemic (TheQuestSage.com, Sl 37) — The companion piece on sleep health, directly relevant to this article’s discussion of orthosomnia and sleep-tracker anxiety.
• Yoga Nidra and the Science of Sleep (TheQuestSage.com, Sl 36) — A companion piece on a non-technological approach to sleep quality, offering a useful contrast to wearable-based sleep tracking.

---