Research team develops a standardized approach to summarize wearable data for research and clinical practise

Smartwatches, sensors and other wearable devices make it possible to continuously monitor aspects of our health and everyday activity. But this capability raises a challenge: how can thousands of measurements collected over days, weeks, or years be transformed into indicators that can be reliably interpreted and compared?

A study led by the Barcelona Institute for Global Health (ISGlobal), a centre supported by the ”la Caixa” Foundation, and the University of Basel proposes a new framework to standardise this process. The method has been developed to be applicable to different types of digital health data and was tested using walking activity and gait data from 2,181 people. The study has been published in npj Digital Medicine. 

Digital devices can record health parameters at a frequency and over periods of time that would have been difficult to achieve only a few years ago. However, having more data does not necessarily mean having better information. To use these data in research or clinical practice, it is necessary to decide which parts are relevant and how they should be summarised. These decisions can influence how the results are subsequently interpreted.

There are currently no internationally accepted guidelines standardising how continuous, high-resolution digital health data should be aggregated, nor is there consensus on the terminology used to describe this process. A previous review cited in the study found, for example, that more than half of the studies using mobile devices that were analysed did not specify which data restrictions they had applied.

A three-step process

To address this problem, the research team used a structured expert consensus process, in which 16 specialists from different disciplines went through several rounds of proposals, assessment and discussion until agreement was reached. The panel included experts in sensor and algorithm development and evaluation, biomechanics, epidemiology, statistics and different clinical fields.

“The framework we propose follows three stages. First, we identify possible ways of summarising the data, taking into account their technical validity, clinical interpretability, sample size and distribution. We then statistically assess how stable the resulting measures are within the same individual and whether they can distinguish between relevant groups - for example between people with and without walking disabilities. Finally, experts compare these results with existing evidence, available international guidelines and practical considerations to select the most appropriate measures,” explains Sarah Koch, first author of the study and a researcher at the University of Basel and ISGlobal.

More than 2,000 people monitored for one week

To test how the framework works, the researchers applied it to data from the European Mobilise-D project. The analysis included 2,181 people, including participants with chronic obstructive pulmonary disease (COPD), multiple sclerosis, Parkinson’s disease, proximal femur fracture and congestive heart failure, as well as healthy controls. Participants wore a sensor for seven days that recorded their mobility in real-world conditions.

The team started with seven measures related to walking activity and gait —including step count, walking speed, stride length, cadence and walking bout duration— and assessed different ways of aggregating them. The process ultimately enabled the selection of 24 mobility indicators, aggregated at daily and weekly levels and related both to the amount and patterns of walking activity and to gait characteristics.

“Even when we applied the method to populations with very different mobility impairments, we obtained clinically interpretable and statistically stable indicators that can be used to compare different individuals or the same individual over time. Moreover, the framework is not designed solely to analyse mobility: its structure can be adapted to other health parameters recorded continuously over extended periods, such as heart rate or temperature,” says Judith Garcia-Aymerich, Deputy Director of ISGlobal and last author of the study.

Reference

Koch S, Buekers J, Alvarez P, Becker C, Bonci T, Carsin AE, Caulfield B, Cobo I, Del Din S, Demeyer H, Frei A, Hausdorff JM, Ionescu A, Aminian K, Klenk J, Lemos-Portela J, Marchena J, Maetzler W, Micó-Amigo E, Piraino P, Singleton D, Sverdlov O, Troosters T, Rochester L, Garcia-Aymerich J. Aggregation framework for continuous digital health measures and application to walking activity data. npj Digit. Med. (2026). https://doi.org/10.1038/s41746-026-03222-z