Forem: Shradha Puri

Understanding Wearable Sensors: What Data You Can Trust

Shradha Puri — Thu, 16 Apr 2026 09:19:08 +0000

After having worn my smartwatch and smart ring nearly every day over the last couple of years, I have found that there is a good balance between appreciating the data on my wrist and viewing it with some healthy skepticism. For example, one morning, my watch will show that I have slept “excellently” and burned a lot of calories from simply taking a walk. The following morning, I feel exhausted despite having recovered “perfectly.”

Sounds familiar?

Most of us bought these devices hoping for accurate readings about our steps, heart rate, sleep and recovery. However, not all sensors give similar results. Some readings are quite good in terms of their accuracy, but others can be described as mere approximations. Based on current research to determine the validity and reliability of these devices as compared to laboratory testing equipment (ECG chest strap, sleep tests), here is a breakdown of what you can actually trust and what data is best taken with a grain of salt.

How Wearable Sensors Actually Work

Most wearable devices rely on a small set of sensors that are the core of all the data we get from our devices.

PPG (Optical Heart Rate Sensors)- PPGs use green, red/infrared light to measure variations in blood volume underneath your skin. These provide HR, RHR, HRV, blood oxygen and recovery data.
Accelerometers and Gyroscopes (Motion Sensors)- These measure movement to track the number of steps, estimate the intensity of activity and differentiate between sleep and awake stages.
Additional Sensors- These include sensors like skin temperature, ECG, EDA or ambient light to detect cycle phases, stress, onset of an illness, readiness score or any more advanced features.

The raw data is then analyzed through an algorithm that produces the graphs and score values we use. The device is remarkable for its small size and the amount of information it packs, but many factors in the real world can affect the accuracy of data.

Heart Rate: Often Reliable, But Not During Chaos

For resting and light daily activity, PPG readings of heart rate from the wrist-worn trackers have proven to be accurate. These usually had an average error of less than 1 BPM in resting conditions. The Apple Watch has consistently performed better and has excellent correlation coefficients; Garmin and other Fitbit trackers also provide reliable HR estimates.

However, for high-intensity exercises, HIIT, resistance training and wrist-intensive activities such as cycling or typing, the accuracy of the PPG measurements significantly drops. Errors increase and many trackers have an absolute deviation of 10-20%+. Multiwavelength sensors have helped mitigate the problem, although recent studies have found that motion artifacts are more influential than skin color differences.

Trust or not: Should be trusted for resting HR and steady state cardio. Best paired with a chest strap for an accurate high-motion workout reading, especially if you’re training for a marathon or are an athlete.

Step Counting: One of the Most Dependable Metrics

Step counting appears to be quite reliable, particularly when walking and running occur under lab conditions. Garmin and Apple Watch can obtain less than 5% accuracy during experiments.

In their “free-living” version, the percentage of inaccuracy amounts to** 6-18%**. Walking, running, pushing a stroller, carrying things and cycling are harder to detect accurately. Since the Oura Ring is finger-based, it may count differently compared to a wrist tracker such as a WHOOP or an Apple Watch.

Trust or not: Very high. Relatively low in the case of non-standard days. For me, it was quite motivating despite inaccuracies in step counts.

Calorie Burn or Energy Expenditure: Treat as a Rough Guide

Calories burned are always a poor indicator in wearable tech. Research indicates mistakes of 20% to 50% or higher since calories cannot be determined directly but must be approximated based on factors such as heartbeat, activity levels, age, body mass index and general algorithms without knowledge of the individual’s metabolism, physical condition and muscle efficiency.

While the Apple Watch usually fares better than most, it still misses by several hundred calories on occasion.

Trust or not: Low when making calculations for everyday life decisions, such as calorie intake.

Sleep Tracking: Strong on Big Picture, Weaker on Details

Sleep/wake detection is another area where wearables, especially smart rings excel. Among various wearables tested by independent researchers, the Oura Ring always scores the highest accuracy (agreement with polysomnography). In one of the studies, the Oura ring 4 demonstrated an excellent Cohen’s kappa of 0.65 for its four-stage analysis (0.60 for Apple Watch, 0.55 for Fitbit), as well as high deep sleep sensitivity (~79.5%).

Yet, all wrist wearables usually overestimate sleep duration and efficiency (by 10-15%), not to mention relatively low accuracy in sleep stage distribution (REM, light, deep).

Trust or not: High for general duration, consistency levels and sleep/wake state. Moderate for a specific percentage per sleep cycle stage. Your “readiness” score for each night only truly comes into context after understanding how it correlates to how you feel the next day.

HRV, SpO2 and Other Metrics

Heart Rate Variability (HRV) and Recovery:

Oura devices excel in this category, particularly at night, where they achieve a high concordance correlation coefficient of up to 0.99 and very low error rates (about 6%) compared to ECGs in 2025 trials.

Blood Oxygen Saturation (SpO2):

Acceptable at rest but susceptible to dropouts when exercising or at high altitudes.

Skin Temperature:

Useful for observing relative fluctuations (disease, menstrual cycle changes) rather than absolute values.

Real-Life Factors That Affect Accuracy

Problems include loose fitting, sweat, tattoos, positioning of the wrist during exercise and the type of movement. Motion artifacts continue to be the problem despite the efforts by manufacturers to minimize errors due to skin tone differences in modern sensors. The wearable device never receives calibration in a laboratory setting.

Simple Habits That Make My Data More Reliable

Here are some simple techniques that will be useful to you without any additional gadgets:

Give preference to trends over daily fluctuations – An occasional spike is not important, but it is a different story when the values and trends are elevated for two weeks.
Add your own notes – Log information about the performance level of your training (on a scale from 1 to 10), energy level or actual sleeping and waking hours, what factors affected them, traveling, etc. The application calculates new baselines based on this information.
Combine sources – Utilize the phone GPS for walking outside, while the cheapest chest strap will measure heart rate for intense workouts. Multiple sensors' data is automatically combined by many health applications.
Perform quick reality checks – Twice a year, cross-check readings against your chest strap at least once in the context of exercise or create a manual sleeping schedule for several days.
Use the right gadget for the purpose you want to solve – Oura ring is perfect for monitoring sleep and recovery, whereas Garmin or Apple watches are better for training and tracking your distance.

The Bottom Line: Smart Use Beats Perfect Hardware

None of these consumer wearables qualify as medical devices, nor will their next-gen versions be any different. The accuracy of heart rate measurements while awake, step count, general sleep tendencies and overnight HRV values can be relied upon. Calorie expenditure and sleep stage detection can be useful but still fall into a "helpful indicator" category, not perfect measurements.

If you use the data in tandem with how you actually feel, that would be the real gold mine. The use of consumer wearables has been highly motivational for me. For example, my Ultrahuman Ring has helped me realize my late-night snacking habits were affecting my heart rate drop during sleep, something which I previously did not pay much heed to.

Do you think some of the metrics in your wearable are more or less accurate than others? What kind of wearable do you have and what makes you believe its information? Let me know in the comments below.

Why Wearable Data Doesn’t Match Reality (And What to Do About It)

Shradha Puri — Tue, 14 Apr 2026 08:09:42 +0000

You put on your smartwatch. You kill your workout and check the results: 12,800 steps walked, 490 calories torched, a great score on sleep, nice heart rate zones. It all sounds accurate. Invigorating, even. Except it’s not. Not really.

Having spent many years with various health tech products that integrate with the likes of Apple Watch, Garmin, Fitbit, WHOOP and Oura, I have noticed an exact same pattern over and over again. Everything appears so clear and clean on the dashboard, but it doesn't always match what was going on in real life. Misleading insights are shared by coaches and bad apps get shipped.

And it’s not just some sporadic noise in the system, it’s an inherent problem every wearable product developer needs to know about. In today’s article, I will talk about why wearable data is often off and what you developers can do about it.

The Promise vs. The Reality: Hard Numbers from Studies

Wearables are sold as precision health tools, but independent validation tells a more sobering story:

Heart rate accuracy: Optical PPG sensors perform reasonably at rest (often within ±3-5 bpm), but errors increase significantly during intense or high-motion activities. Active heart rate accuracy ranges from around 67-86% depending on the device, with Apple Watch generally leading at ~86% and others like Garmin and Fitbit lower during dynamic movement. Dark skin types tend to have more inaccuracies due to the absorption of the light-green light used in most sensors by the melanin in their skin.
Step count: Moderate accuracy at about 68-82% overall. They generally underestimate by about 8-12% overall in free-living conditions and higher in non-ambulatory movements such as cycling, stair climbing, and load carriage. Garmin usually performs slightly better under controlled conditions, with the opposite occurring otherwise.
Energy expenditure: Energy expenditure is the least accurate among all metrics. Inaccuracies in energy expenditure often exceed 25-28%, with an accuracy rate of around 56-71%. The Apple Watch seems to have a higher accuracy rate for heart rate at 86.31% and 71.02% for energy expenditure. While Garmin is most accurate for tracking step count at an 82.58% accuracy. Estimates are based on indirect equations that use heart rate, movement data, age, weight and model-based formulas.
Sleep metrics: Devices seem to perform quite well when it comes to detecting sleep vs. wake state, as sensitivities are often greater than 90%. They, however, overestimate sleep time and efficiency, especially efficiency. Accuracy when classifying sleep stages is moderate, varying widely by device, with the Oura Ring being recommended over wrist devices.

Such biases are not uncommon, one-off incidents but rather systematic issues that impact millions of individuals everyday. When you design applications for fitness training, workplace health programs, insurance underwriting, or medical research, your designs stand on much shakier ground than you realize.

Why Wearable Data Diverges from Reality

This discrepancy arises due to various factors:

1. Technical Limitations
First of all, consumer wearables typically mount onto your wrist. They use 3D accelerometers along with optical heart rate sensors. The latter type is highly vulnerable to motion artifacts. Sweat, improper wear, tattoos, hair, skin color discrepancies and even applying lotion affects sensor accuracy. Any intense movement causes displacement of the watch against your skin.

2. Algorithm Assumptions
The algorithm is trained using population data. However, your unique physiological features (VO2 max, muscle fiber ratio, basal metabolic rate, medications, etc.) heavily impact "true" values. Unless you provide and continuously update these metrics, there’s no reliable means of compensation for the device.

3. Inherent Biases
A series of academic papers has discussed reduced accuracy for individuals with darker skin tone, larger wrists and peculiar body compositions. This problem doesn't come from the software implementation but from historical biases in training sets.

4. Lack of Continuous Calibration
In a lab, researchers use ECG chest straps and metabolic carts for validation. On your wrist during daily life? The device is making educated guesses with no continuous calibration against medical-grade equipment. It’s flying somewhat blind.

5. Data Pipeline Issues
Even when the hardware captures something useful, the way data is sampled, aggregated, filtered and synced to the cloud introduces further gaps and delays.

My Personal Reality Check

Since I have been tracking my data for years, I now know that one must be skeptical about these figures. There were days when I had supposedly slept really well, according to my smartwatch, but I felt sleepy and realized that I had several moments where I woke up throughout the night. Or days when my smartwatch displayed amazing calories burned, even though I only had a relatively light gym session.

What this has taught me is that one must be able to think differently regarding the data coming from the device. Your feelings, performance and personal perception will always come first. It would not do well to blindly trust everything shown by your wearable because it might make you overwork yourself or ignore your bodily needs.

Practical Ways Regular Users Can Bridge the Gap

It doesn’t require any technical knowledge to boost the accuracy of your results. Here are some techniques that have proved to be most effective for myself and many others:

Cross-Check and Manually Adjust

Periodically compare your device’s measurements with your real-life perceptions. If you notice something suspicious about your sleep score, add notes to the application and make manual corrections whenever you can. This will help you find patterns over time.

Add Your Own Context

Almost all popular apps offer the ability to enter additional data. For example, the type of workout, Rate of Perceived Exertion, current mood, nutrition or sickness. Use those functions. They can provide much more accurate calculations for your personal situation.

Combine Multiple Sources

Avoid trusting only wrist measurements. You can use GPS tracking on your smartphone for outside jogging or walking to get a more precise number of steps and covered distance. And in cases of workouts while training for a marathon, you may want to employ a heart rate monitor for periodic checking. Most of the applications gather information from various sources.

Look for Transparency

Use devices and applications that have confidence measures and ranges if possible ("calories ±20%", for instance). Study the limitations mentioned by the producers themselves in the documentation. Once you know your weak sides, the information will be much more valuable.

Periodic Reality Checks

Do a simple verification day every few months when comparing your watch with more precise devices, such as a chest strap while doing sports or logging all the details about your sleep manually. Take the discrepancies into account when analyzing your data and decide how much to trust it.

Focus on Trends, Not Single Days

One inaccurate reading on a certain day does not mean your wearable data is useless. Analyze long-term trends to get a bigger picture.

Choose the Right Tool for Your Needs

If you want some more encouragement to work harder towards your goal, just use any wrist device. In case of a more serious approach to your sleep and recovery or training, smart rings like Oura and chest straps will help.

The Road Ahead

Although newer models of wearables are getting better with more advanced sensors and software, there are still limitations. A perfect degree of wrist-based precision under all circumstances is still impossible.
What matters most is not perfection but proper use. The perfect combination would be the fusion of the quantitative data and the qualitative knowledge of your own body.

Wrap Up

The data obtained from wearables is extremely useful. However, it will only be so when we drop our expectation that it will be perfect. Treat it as one helpful voice in the conversation about your health, not the only voice you rely on. Always pay attention to what your body tells you first, then utilize the device to supplement information received from it.

More things that can help get the most out of your wearable data are eliminating what are clearly inconsistencies, manual entries to add what’s happening in your surroundings and tracking your trends.

The most successful people aren’t necessarily those with the latest high-tech devices. The most successful ones know their limitations and make the data work for them and not the other way around.