Questioning Personalised Health: Limitations of Interpreting Wearable Data at the Individual Level

Article information

Korean J Health Promot. 2026;.kjhp.2026.00255

Publication date (electronic) : 2026 May 8

doi : https://doi.org/10.15384/kjhp.2026.00255

Jun Woo KWON, MEd

Department of Physical Education, Seoul National University, Seoul, Korea

Corresponding author: Jun Woo KWON, MEd Department of Physical Education, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul 08826, Korea Tel: +82-2-880-7788 Fax: +82-2-872-2867 E-mail: bichoncontin@gmail.com

Received 2026 March 29; Revised 2026 April 20; Accepted 2026 April 23.

The rapid adoption of wearable technologies has transformed how individuals engage with their health. Devices that continuously monitor heart rate, physical activity, sleep, and other physiological parameters are now widely promoted as tools for personalised health management. This shift aligns with broader trends in digital medicine, where data-driven approaches promise to tailor interventions to individual needs. This broader shift reflects the ongoing transformation of healthcare through digital medicine, characterised by increasing reliance on data-intensive approaches and continuous monitoring outside traditional clinical settings [1]. However, the increasing reliance on wearable-derived metrics raises an important question: do these data truly represent individualised health information, or do they merely create the appearance of personalisation? This issue is particularly relevant in the context of health promotion, where wearable-derived data are increasingly used to guide individual behaviours and decisions. While such data may enhance awareness and support engagement with health-related activities, their interpretation is not always straightforward. Variability in measurement, differences in usage contexts, and reliance on population-based algorithms may limit the extent to which these data reflect individual health states. As a result, the apparent precision of wearable outputs may not necessarily correspond to meaningful or actionable insights.

Wearable devices are often presented as empowering tools that enable individuals to take control of their health. Their appeal lies in the ability to generate continuous, real-time data outside traditional clinical settings [2]. This accessibility has positioned wearables as key instruments in health promotion, particularly in encouraging physical activity, improving sleep habits, and facilitating self-monitoring. Recent developments have further expanded their role, with wearable sensors increasingly integrated into predictive models and digital health ecosystems [3,4]. These advances may suggest a future in which health decisions are informed by highly granular, individual-level data, but more data does not necessarily mean better insight.

Despite this promise, the validity of wearable-derived data remains a fundamental concern. Numerous studies have demonstrated variability in the accuracy of commonly measured parameters such as heart rate and energy expenditure, particularly across different populations and activity contexts [5]. Systematic reviews have also highlighted inconsistencies in the reliability and validity of commercially available devices, suggesting that measurement error is not negligible [6]. While such limitations may be acceptable for general tracking purposes, their implications become more significant when these data are interpreted as clinically meaningful indicators of individual health.

Beyond measurement accuracy, a more critical issue may lie in how these data are interpreted. Wearable technologies now generate large volumes of personal health data. However, translating these data into clinically meaningful or actionable insights remains uncertain, particularly given ongoing challenges in the definition and validation of digital biomarkers [7,8]. Unlike conventional physiological or biochemical biomarkers, which are typically measured under controlled conditions and supported by established validation frameworks, digital biomarkers derived from wearable devices often rely on heterogeneous data sources and less established validation approaches [7]. As a result, their interpretation at the individual level may be inherently constrained. Without appropriate contextualisation, such data may be misinterpreted. For example, heart rate or sleep-related metrics obtained from wearable devices may vary depending on activity type, measurement conditions, or device characteristics, reflecting known variability in wearable-derived measurements [5,6]. Such fluctuations may be influenced by transient physiological or environmental factors rather than meaningful changes in underlying health status. In this sense, the assumption that more data necessarily leads to better insight may be overly simplistic.

This challenge becomes more apparent in real-world settings. Continuous monitoring can certainly increase data availability, but it does not automatically improve interpretability. In practice, the clinical use of these data is still limited by the lack of standardised validation approaches and the absence of universally accepted gold standards for many digital biomarkers [8]. In addition, linking digital signals to clinically meaningful outcomes in complex, real-world environments remains a recognised challenge [8]. As a result, numerical outputs generated by wearable devices—such as composite indices or summary scores—may convey a sense of precision that is not fully supported by their methodological foundations. Caution may therefore be warranted when interpreting such metrics as definitive indicators of individual health status.

Building on these considerations, the concept of the illusion of personalisation is proposed. This concept describes the gap between the apparent individual specificity of wearable-derived data and their limited interpretability at the individual level. The illusion may arise from several factors. Continuous data streams, numerical outputs, and algorithmically derived summaries can create a strong impression of precision. At the same time, these data are shaped by measurement variability, contextual influences, and population-based modelling. Together, these factors may constrain their validity when applied to individuals. As a result, wearable-derived metrics can appear highly personalised, while in practice they may reflect broader patterns rather than precise individual states. This distinction between perceived and actual personalisation is particularly relevant in health promotion and preventive care, where such data are increasingly used to guide individual decision-making. In practice, wearable-derived data may be most appropriately used to monitor general trends or support behaviour change rather than to inform precise individual-level decisions. For example, gradual changes in activity patterns or sleep duration may provide useful signals for health promotion. However, interpreting short-term fluctuations or relying on single measurements as indicators of individual health status may be more problematic. In such cases, these data may require cautious interpretation, particularly when used outside validated clinical contexts. Recognising this distinction may help ensure that wearable technologies are used in ways that are both meaningful and appropriate for individual health management.

A central issue underpinning these concerns is the difficulty of translating population-level data into individual-level conclusions. Much of the evidence supporting wearable-derived metrics is based on aggregated data, which may not generalise reliably to individual users. In clinical research, it is well recognised that treatment effects and risk predictions derived from populations do not necessarily apply uniformly to individuals [9]. This limitation extends to wearable technologies, where algorithms trained on population data are often used to generate personalised outputs. The apparent individualisation of these outputs may therefore mask substantial uncertainty. Population-level evidence does not guarantee individual-level validity. What works on average does not always work for the individual.

Recent advances in precision medicine and digital health have sought to address this challenge by incorporating more sophisticated modelling techniques. Approaches such as dynamic predictive modelling and digital twins aim to capture individual variability more accurately [4,10]. However, these approaches also highlight the complexity of achieving true personalisation in practice. Developing reliable individual-level models requires rigorous validation. It also demands careful consideration of uncertainty, both of which remain ongoing challenges in the field [10]. In this context, the expectation that wearable-derived data alone can provide precise, individualised health insights may be premature.

Importantly, the issue is not that wearable technologies lack value. On the contrary, they offer significant opportunities for enhancing health awareness, supporting behaviour change, and facilitating large-scale data collection. Their utility in health promotion is well established, particularly in encouraging physical activity and monitoring general health trends [2,3]. However, the growing perception of wearables as tools for precise, individualised health assessment warrants careful scrutiny.

In the context of health promotion and preventive medicine, this distinction is particularly relevant. The use of wearable data to guide individual health decisions—such as adjusting exercise intensity, modifying sleep habits, or interpreting physiological fluctuations—requires a clear understanding of the limitations of these data. Without such understanding, there is a risk that individuals may attribute undue significance to metrics that are inherently uncertain or context-dependent.

A more cautious approach to interpreting wearable-derived data is needed. Rather than assuming that continuous monitoring equates to personalised insight, it may be more appropriate to view these data as indicative of general trends rather than precise individual states. Emphasising the limitations of wearable metrics does not diminish their value; instead, it promotes a more realistic understanding of their role in health management.

Ultimately, the promise of personalised health through wearable technologies must be balanced against the realities of data accuracy, interpretation, and individual variability. Personalised does not always mean precise. While wearables have the potential to contribute meaningfully to health promotion, their outputs should not be uncritically regarded as definitive representations of individual health.

In conclusion, wearable health data may appear to offer personalised insights, but their interpretation at the individual level remains uncertain. As the use of these technologies continues to expand, it is important to critically evaluate the assumptions underlying their application. In this regard, wearable-derived data may be most appropriately used to support general health awareness and behavioural guidance rather than as standalone indicators of individual health status. Integrating these data with clinical context and professional judgement may be important for their appropriate use in practice. In practical terms, this may involve considering wearable-derived data alongside clinical assessment, patient-reported symptoms, and contextual factors rather than interpreting these metrics in isolation. Such an approach may help reduce the risk of overinterpretation while preserving the potential value of wearable technologies in supporting health-related decision-making. A more nuanced understanding of wearable data—one that acknowledges both their potential and their limitations—will be essential for advancing their role in health promotion and preventive care.