Wearable Technology and Biohacking

Wearable technology includes watches, rings, patches, chest straps and other devices that collect signals from the body or its movement. The quantified-self approach uses those observations to look for personal patterns. Biohacking is a much broader and less precise label: it can mean adjusting a bedtime after reviewing a sleep log, but it is also used for supplements, extreme environmental exposure, implants and unapproved biological experiments.

Those activities should not be treated as one continuous path toward “optimisation.” More intervention does not necessarily produce more benefit. A useful first step is to separate them by reversibility, uncertainty and potential harm.

A wearable captures a physical signal and converts it through several layers of processing. Each layer can add useful information—and another source of uncertainty.

1. Body eventThe heart beats, the wrist moves, skin temperature shifts or glucose enters interstitial fluid.
2. Sensor signalLight, electricity, motion or chemistry is sampled. Fit and context affect quality.
3. ProcessingSoftware removes noise, fills gaps and rejects suspect readings.
4. EstimateData becomes steps, sleep stages, energy use or another labelled output.
5. InterpretationThe app compares a result with a personal or population baseline.
6. DecisionThe user rests, trains, worries, changes a habit or seeks assessment.

Photoplethysmography, or PPG, illustrates the chain. Green or infrared light sent into the skin changes as blood volume pulses. The sensor records reflected light; software then estimates pulse timing and may derive heart rate, pulse-rate variability, oxygen saturation or rhythm alerts.¹ The device does not directly see “stress,” “recovery” or cardiovascular health.

An accelerometer records motion, but steps, activity type, sleep and calories are model outputs. A temperature sensor usually measures local skin temperature, not core temperature. A continuous glucose monitor is a separate minimally invasive sensor that samples interstitial glucose; a watch may display its data without measuring glucose itself.

Consumer sensors are most valuable when the decision stays within the strength of the evidence. Trends are often more informative than isolated readings, especially when the same device is used under similar conditions.

Correlation is not agreement. Two devices can rise and fall together while differing enough to change a decision. Likewise, a small average error across a group is not a guarantee about one person’s reading. Look for individual limits of agreement, failure rates and subgroup performance—not only a correlation coefficient or headline accuracy.

Technical testing asks whether a sensor can capture a signal under controlled conditions. Clinical validation compares a feature with an appropriate reference method. Free-living testing asks whether performance survives movement, imperfect fit, changing environments and ordinary use. Evidence at one level does not automatically establish the next.

Look beyond a headline percentage. Was the study independent? How many people participated? Did they resemble intended users? Was the feature tested during daily life or only while participants sat still? An average can hide large errors for particular people, activities or ranges. Optical wearables need testing across perfusion, motion and real-world conditions.¹³ For medical pulse oximeters specifically, FDA has proposed updated performance and labelling recommendations intended to improve accuracy across skin tones.⁹ These issues are device- and metric-specific, so blanket claims in either direction are unhelpful.

A personal baseline does not erase design bias

Comparing someone with their own history can reduce misleading comparisons with population averages, but it cannot recover a signal the hardware captured poorly. Validation samples should include varied skin tones, ages, body sizes, health conditions, movement patterns and assistive-device use. Comfort and accessibility matter too: a sensor that cannot be worn consistently, an interface that cannot be read or a subscription that interrupts access will perform differently outside the study. Missing data are not neutral if they occur most often during the activity or symptom a person hoped to understand.

FDA’s 2026 general-wellness guidance describes when certain low-risk products may sit outside active medical-device enforcement.¹⁹ It is not a universal certificate of accuracy. Conversely, a wellness label does not prove a product is poor; it tells the reader not to assume it was reviewed for diagnosis or treatment. Firmware can change after a study, so the tested model and software version matter.

Screening signals still need confirmation

Large smartwatch studies show that passive optical algorithms can identify pulse patterns worth investigating.⁶⁷ Yet an irregular-rhythm notification is not a diagnosis. Charging gaps, motion, inconclusive recordings and rhythms other than atrial fibrillation complicate interpretation. The useful response is proportionate confirmation through the device’s instructions and appropriate clinical care—not panic, dismissal or self-treatment.

A personal experiment can help answer a modest question about a reversible routine. It cannot remove placebo effects, random variation or bias, and it does not make a risky intervention safe.

1. Ask a meaningful question

   “Does a steadier bedtime help me feel more alert?” is better than “How can I maximize every score?”

2. Observe normal variation

   Collect a baseline before assigning meaning to one unusually good or bad day.

3. Change one low-risk factor

   Keeping other routines reasonably stable makes the result easier to interpret.

4. Choose outcomes in advance

   Include energy, comfort or performance—not only the app’s composite score.

5. Note obvious confounders

   Illness, travel, stress, alcohol, medication changes and schedules can dominate the signal.

6. Set a stop rule

   Stop if the experiment causes pain, marked distress, unsafe fatigue, restrictive eating or another meaningful adverse effect.

Formal N-of-1 studies can use repeated comparisons, randomisation and other safeguards; casual self-tracking rarely does.²⁵ A result may still be personally useful, but it should be described honestly: “Under these conditions, I noticed this pattern.” It does not show that the change caused the outcome or that other people should copy it.

Use a decision journal, not just a graph

Record what you planned before seeing the result, the outcome that mattered, relevant context and what would change your mind. This reduces the temptation to invent an explanation after every fluctuation. Review at a chosen interval instead of reacting to each notification. If an effect vanishes when routine life returns, that is useful evidence too.

Keep the intervention proportionate to the uncertainty. Moving a walk or bedtime is different from changing medication, insulin, supplements, food groups or exposure to extreme temperature. Medical treatment and high-risk practices should not be adjusted on the authority of a consumer score. Symptoms, safety instructions and professional assessment sit above a self-experiment.

A bedtime log, a concentrated compound and an implanted device are not simply stronger versions of the same idea. Judge any “hack” by what enters or alters the body, how reversible it is, what evidence supports it and what happens if the assumption is wrong.

Dietary supplements are regulated, but in the United States they generally do not undergo drug-style premarket approval for safety and effectiveness.²⁰²¹ Ingredients can interact with medicines, vary between products or be inappropriate during pregnancy, illness or surgery. Genetic associations and consumer “nutrigenomic” reports do not turn that uncertainty into an individual prescription.

Environmental stressors need the same restraint. A 2025 review of cold-water immersion found a mixed, time-dependent evidence base—not a universal duration, temperature or guaranteed recovery effect.²⁴ A tracker can document a response; it cannot show that the exposure is safe for a particular person.

Passive NFC or RFID implants do not contain GPS merely because they can transmit stored data to a nearby reader. Implantation still creates questions about infection, tissue response, migration, removal, access control and device compatibility. Published evidence is too sparse to promise a universal complication rate.²³ DIY gene therapy goes further: FDA specifically warns that self-administered gene-therapy products are unapproved and may be dangerous.²²

Continuous tracking can produce a sensitive record of sleep, location, routines, fertility, illness, exercise and social patterns. The device on the body is only the first stop. Data may move through a phone, an account, cloud infrastructure, analytics providers and authorised partners.

Read the privacy notice as an operating map. Ask what is collected automatically, whether processing happens on-device or in the cloud, how long raw and derived data are retained, who receives them, whether research use is opt-in, and how to export and delete an account. Deletion may not immediately remove lawful backups or already de-identified and aggregated information, so inspect the company’s exact promise rather than assuming a universal outcome.

Legal protection depends on who holds the data and where. In the United States, HIPAA does not automatically cover information held only by a consumer app outside a covered healthcare relationship; the FTC’s Health Breach Notification Rule may still apply.²⁹³⁰ In the European Union, health data receive special protection under the GDPR, but rights and lawful uses still depend on context.²⁸

AI coaching adds another inference layer

An assistant can summarise a long log, surface possible patterns and help prepare questions. It can also invent explanations, miss contraindications and sound certain about incomplete inputs. Model behaviour and vendor algorithms can change. Do not upload more health information than needed, and do not let a conversational answer diagnose symptoms or override device instructions and qualified care.

Attention is part of the intervention. A tracker can support a habit, but repeated checking can make an uncertain estimate feel urgent. Clinicians coined orthosomnia to describe cases in which people became preoccupied with perfecting wearable sleep data and, paradoxically, made sleep more difficult.²⁶ Later research suggests that maladaptive sleep-tracking behaviours exist, while prevalence and definitions remain unsettled.²⁷

A tracking pause is not failure; it is a test of whether the tool remains useful. If data anxiety persists, restrictive behaviour develops or symptoms are being managed through repeated measurement rather than appropriate care, step away from the dashboard and discuss the pattern with a qualified professional.

Marketing often moves from “the sensor detected a signal” to “the product improves your health” without testing the steps in between. These questions reveal the missing links.

1. What is directly sensed?Separate light, motion, temperature or chemistry from the label produced by software.
2. Which exact version was tested?Model, placement, firmware and app version can all matter.
3. What was the reference?A sleep diary, another watch and laboratory polysomnography answer different questions.
4. Who and where were studied?A small resting sample may not generalise to daily life, intense exercise or a different population.
5. Who funded and analysed it?Independence, preregistration, attrition and conflicts help readers judge reliability.
6. Is the outcome meaningful?Better agreement or a changed score does not necessarily mean better health or function.
7. What exactly was authorised?Check the feature, intended use, population, jurisdiction and current software—not the brand halo.
8. What decision remains outside scope?Ask what the result cannot diagnose, exclude or safely change.

Choose the device from the question backwards

Define the decision first. Then look for independent validation of that metric, a usable export, clear privacy controls, comfort, battery life, accessibility, support lifespan and total cost including subscriptions. Decide what would make you stop tracking before the novelty begins.

Wearables are unusually good at making invisible routines visible. A stable resting trend, a record of sleep opportunity or a prompt to move can reveal something that memory alone would miss. Their weakness is the smooth transition from noisy signal to confident story.

The safest and most useful approach is deliberately modest: understand what was sensed, learn how the estimate was validated, keep the decision beneath its evidence ceiling and compare the dashboard with lived function. Test one reversible change at a time. Protect the data trail. Treat medical alerts as prompts for appropriate confirmation, not invitations to diagnose or dose yourself.

Human intelligence is not located in a readiness ring. Technology can support it indirectly—by helping people notice patterns, ask better questions and prepare better conversations. The goal is not a frictionless body or a perfect graph. It is informed agency: knowing when a number deserves attention, when it needs confirmation and when it deserves to be ignored.

1. Photoplethysmography and cardiovascular monitoring. Peer-reviewed review, PMC.
2. Keeping Pace with Wearables. Living umbrella review of measurement accuracy, 2024.
3. Investigating sources of inaccuracy in wearable optical heart-rate sensors. npj Digital Medicine study.
4. Wearable devices for heart-rate variability. Systematic review.
5. Validation of nocturnal resting heart rate and HRV in consumer wearables. Small device-comparison study, 2025.
6. Large-scale assessment of a smartwatch irregular-pulse notification. Apple Heart Study.
7. Wearable detection of atrial fibrillation. Fitbit Heart Study.
8. Pulse oximeter basics and limitations. US Food and Drug Administration.
9. Improving pulse-oximeter performance across skin tones. FDA recommendations.
10. Cuffless Devices for the Measurement of Blood Pressure. American Heart Association scientific statement.
11. Six consumer sleep trackers compared with polysomnography. Independent 2025 study.
12. Consumer Sleep Technology. American Academy of Sleep Medicine position statement, 2018.
13. Accuracy and precision of energy expenditure, heart rate and steps measured by combined-sensing Fitbits. Systematic review and meta-analysis.
14. Wearable estimates of cardiorespiratory fitness. Systematic review and meta-analysis.
15. Do not use watches or rings that claim to measure blood glucose. FDA safety communication.
16. First over-the-counter continuous glucose monitor. FDA clearance announcement.
17. The use of continuous glucose monitors in people not living with diabetes. Narrative review, 2024.
18. Expert Clinical Interpretation of CGM Reports From Individuals Without Diabetes. Expert study, 2025.
19. General Wellness: Policy for Low Risk Devices. FDA guidance, January 2026.
20. What “FDA approved” means. FDA consumer guidance.
21. FDA 101: Dietary supplements. FDA consumer guidance.
22. Information about self-administration of gene therapy. FDA safety information.
23. Biohacking and Chip Implantation in the Human Hand. Peer-reviewed clinical review.
24. Effects of cold-water immersion on health and wellbeing. Systematic review and meta-analysis, 2025.
25. N-of-1 comparative studies. UK government research guidance.
26. Orthosomnia: Are Some Patients Taking the Quantified Self Too Far?. Three clinical examples.
27. Prevalence of Orthosomnia in a General Population Sample. Cross-sectional study using a proposed definition.
28. General Data Protection Regulation. Official EU text.
29. Consumer health information: HIPAA, the FTC Act and breach notification. US Department of Health and Human Services.
30. Complying with the Health Breach Notification Rule. US Federal Trade Commission.