What a Fitness Tracker Won’t Show You, Even After Years of Wear, and What a DNA Test Adds

Fitness trackers, smartwatches, Whoop bands and Oura rings — all these wearables collect data on your sleep, pulse, activity and recovery every day. In the morning the app shows how long you slept, how your heart rate changed and how the algorithm rated your sleep quality, your level of strain or your body’s readiness for a new day. The data build up over months and years, the charts grow ever more detailed, but a question remains: does this mean we already know enough about our own body?

The data from smart devices let you track changes over the course of a day, a month or a year. Yet the figures and their dynamics on their own do not show all the factors that affect the body’s physiology. It is shaped by current health, lifestyle, environmental conditions and individual biological features, including genetic ones.

What a Fitness Tracker Measures

Wearables are increasingly used for continuous physiological monitoring both in clinical settings and at home (Babu et al., Annu. Rev. Med., 2024). Their built-in sensors capture primary physiological signals. A photoplethysmograph records pulse-driven changes in blood volume in the microvascular bed, an accelerometer records wrist movement, and a temperature sensor records skin temperature.

From these signals, algorithms calculate heart rate, sleep duration, heart rate variability and recovery scores. Some of the indicators in the app are not measured directly but are the result of data processing. Their accuracy therefore depends on the characteristics of the sensors and algorithms, the position of the device on the body and the conditions of wear.

A meta-analysis of 24 systematic reviews, summarising 249 unique validation studies and covering 430,465 participants, confirms considerable heterogeneity in accuracy between devices and metrics: the mean bias for heart rate was about ±3%, whereas the error for physical-activity intensity ranged from 29% to 80% depending on the level of exertion (Doherty et al., Sports Medicine, 2024).

Long-term observation helps reveal individual dynamics but does not remove measurement errors. Random deviations may partly average out, whereas a systematic bias of a sensor or algorithm usually persists. The comparability of results can also change after a software update, a change of device or a change in the conditions of wear.

Why Tracker Data Is Called a Digital Phenotype

Despite the errors of individual measurements, data accumulated over time help determine a typical range of indicators and notice when sleep, pulse or activity deviate from the level usual for a particular person. They also make it possible to trace the conditions under which such changes arise — for example, after sleep deprivation, intense physical exertion or a change in the daily routine.

This digital representation of a person’s physiological and behavioural characteristics is called a digital phenotype in the scientific literature. It includes both the data a person enters themselves — for example, a wellbeing rating or logging a workout manually through the app — and the data the device records automatically, without the user’s involvement or any special action. The indicators from a fitness tracker mostly belong to the second type (Dlima et al., JMIR Bioinformatics and Biotechnology, 2022).

A digital phenotype shows how observed indicators change under real-world conditions. Genetic data belong to a different level of information: they do not explain every single fluctuation, but they can provide context on the inherited features linked to the relevant area.

How Genetic Data Relate to a Digital Phenotype

A genetic test determines inherited DNA variants and uses data from large genome-wide association studies to estimate their combined contribution. For complex polygenic traits, the result shows a relative genetic predisposition compared with a reference population. It does not reflect current health and does not predict how exactly a trait will manifest in a particular person.

Unlike the indicators of a digital phenotype, inherited genetic variants remain constant throughout life, so it is enough to determine them once (Xiang et al., Genome Medicine, 2024). At the same time, their scientific interpretation can be updated along with new GWAS, reference databases and methods of calculation. So it is not the genetic data themselves that change, but the scientific models used to assess their significance.

For more on how a polygenic risk score is formed and what its numbers mean, see What a PRS is, and what the numbers in your report mean.

How Wearable Data Is Used in Genetic Research

Within the ABCD project (Adolescent Brain Cognitive Development), devoted to the cognitive development of the adolescent brain, more than 250 features derived from Fitbit data were analysed and used as quantitative digital phenotypes to search for genetic associations. This approach helped identify 16 genome-wide significant loci and 37 genes linked to the psychiatric phenotypes under study (Liu et al., Cell, 2025).

This shows that genetic data can complement a digital phenotype and help investigate the biological factors linked to differences between people. At the same time, the result of an individual DNA test does not explain a particular tracker reading on a particular day: a genetic predisposition describes a long-term context, while the value on the screen is also shaped by the body’s current state, behavioural factors, environmental conditions and measurement.

The Limits of Personal Data Interpretation

A wearable does not analyse a person’s genotype. It tracks sleep, pulse, movement and other measurable features, but it gives no information about the genetic features linked to differences in these features between people. A genetic test, in turn, does not measure the current level of stress, the quality of sleep on a particular night or the current heart rate variability: these indicators depend on routine, the state of the body, habits, environmental conditions and other factors.

A higher or lower estimate of genetic predisposition does not show how a person slept last night. The specific value of a sleep score, HRV or another tracker indicator depends on many factors that a genetic test does not measure. Neither of the two tools establishes a diagnosis or replaces a doctor’s consultation. These data are of the greatest value when considered together with laboratory indicators, current state and clinical context.

Data from systematic reviews on the validation of wearables show that measurement accuracy differs substantially between models and types of indicator (Doherty et al., Sports Medicine, 2024). So even indicators of the same type obtained from different devices should not be compared directly.

The link between these indicators and genetics should also be interpreted with caution. Population studies can estimate the heritability of individual traits, but such an estimate does not show what part of a particular person’s HRV or sleep score is linked to genetic features.

In practice, tracker data and genetic test data do not replace each other. Genetic data add a long-term context on particular areas, while a tracker shows how the corresponding indicators change in everyday life. Together with data on current state and examination results, this gives a broader context for interpretation.

An Apixmed Prism report can complement tracker data with information about genetic predisposition in the areas included in the chosen test — in particular, sleep, recovery, the response to physical exertion, nutrition or metabolic features. Data from a wearable are not uploaded into the report automatically, but they can help formulate more precise questions about the test results and for discussion with a doctor.

The Value Is Not in the Amount of Data but in Its Interpretation

Data from a wearable reflect measurable features over a certain period and under specific conditions, whereas a genetic test assesses inherited features linked to individual differences in sleep, recovery, metabolism and the response to exertion. Genetic context does not explain every change in tracker indicators, but it adds information about inherited features that cannot be obtained even through long-term observation with a wearable.

Learn more about your genetic context

The results of a genetic test are not a diagnosis and not a substitute for a doctor’s consultation. An Apixmed Prism report provides genetic context that complements the findings of examinations and helps you make decisions together with your doctor.

Sources

1. Liu, J. J., Borsari, B., Li, Y., Liu, S. X., Gao, Y., Xin, X., Lou, S., Jensen, M., Garrido-Martín, D., Verplaetse, T. L., Ash, G., Zhang, J., Girgenti, M. J., Roberts, W., & Gerstein, M. (2025). Digital phenotyping from wearables using AI characterizes psychiatric disorders and identifies genetic associations. Cell, 188(2), 515–529.e15. https://doi.org/10.1016/j.cell.2024.11.012

2. Doherty, C., Baldwin, M., Keogh, A., Caulfield, B., & Argent, R. (2024). Keeping Pace with Wearables: A Living Umbrella Review of Systematic Reviews Evaluating the Accuracy of Consumer Wearable Technologies in Health Measurement. Sports Medicine, 54(11), 2907–2926. https://doi.org/10.1007/s40279-024-02077-2

3. Dlima, S. D., Shevade, S., Menezes, S. R., & Ganju, A. (2022). Digital Phenotyping in Health Using Machine Learning Approaches: Scoping Review. JMIR Bioinformatics and Biotechnology, 3(1), e39618. https://doi.org/10.2196/39618

4. Xiang, R., Kelemen, M., Xu, Y., Harris, L. W., Parkinson, H., Inouye, M., & Lambert, S. A. (2024). Recent advances in polygenic scores: translation, equitability, methods and FAIR tools. Genome Medicine, 16, 33. https://doi.org/10.1186/s13073-024-01304-9

5. Babu, M., Lautman, Z., Lin, X., Sobota, M. H. B., & Snyder, M. P. (2024). Wearable Devices: Implications for Precision Medicine and the Future of Health Care. Annual Review of Medicine, 75, 401–415. https://doi.org/10.1146/annurev-med-052422-020437