The surveillance of physical activity, at a population level, is being met with new demands [1]. Surveillance data should now ideally capture physical activities of all intensities across 24 hours, sedentary behavior, and sleep [1]. These data are very difficult to collect accurately using the traditional self-reported measures widely implemented in physical activity surveillance, primarily because habitual activities and sleep are difficult to recall, and therefore, the estimates reported are often incorrect [2].

In an attempt to overcome the limitations of self-reported measures, sensor-based measurements have been implemented in a few select cases of physical activity surveillance [3-7] and in an increasing number of large cohort studies [8-10]. Although the goals of surveillance and cohort research are inherently different—cohort research informs physical activity guidelines, whereas surveillance monitors adherence to guidelines—the challenges they face in implementing sensor-based measurements are similar. These measurements were only possible at considerable expense, with long delays in producing useful data, and with a considerable burden to all users, both those conducting the data collection and those participating in the studies. Burden can be defined as the direct and indirect financial and resource costs of interacting with the system, including the cost of equipment and expert staff, the time required to recruit participants and coordinate a meeting in person, and the time taken to analyze data and produce useful results. Under this definition, the unacceptable burden of traditional sensor-based methodologies can be summarized quite concisely. The sensors are too expensive, recruitment and data collection require a center and involve considerable logistics for participants and staff, and data analysis requires expert personnel. This burden typically results in a lot of time passing before summary reports and findings based on the collected data reach policy makers. Therefore, we need to develop a methodology for implementing sensor-based measurements of physical activity, sedentary behavior, and sleep, which is lower in burden for all users, in line with the World Health Organization’s global strategy on digital health [11].

To date, we have been largely limited by the available technology. However, recent technological advances could enable the development of new systems for the measurement of physical activity, sedentary behavior, and sleep. Technological advances include a new generation of easily attachable, relatively cheap, discrete sensors capable of secure data transfer via Bluetooth connectivity; new smartphone apps for the integration of sensor and participant input, cloud storage, and automated analysis capacity; and web-based applications for the real-time tracking of data collection progression. These technologies potentially give us the ability to rapidly and automatically analyze data and provide useful, timely feedback to all users in a manner compliant with current data privacy regulations. Additionally, such technologies will increase the potential for collecting and harmonizing data within and across countries in the future, if adapted as a standardized methodology for accelerometry data collection. Thus, exploiting these new technological developments will be the first step toward improving the way we collect accelerometry data at scale.

The surveillance of physical activity, sedentary behavior, and sleep project team (SurPASS) envisions incorporating this new generation of technical solutions and infrastructure in the development of a system for sensor-based surveillance that can be used globally. We see the current protocol as the first step, where we document a method for the development and evaluation of a sensor-based system for surveillance among working age adults at a national level, in Denmark. In this protocol, we will address three aims:

The project started in September 2020. Phase 2 was completed in February 2022 and phase 3 began in March of 2022. Findings will be published in 2023. The proposed system implementation in phase 3 is outlined below (Figure 4).

This protocol describes the developmental process that is the first step toward the next generation of surveillance tools for physical activity, sedentary behavior, and sleep. In designing the protocol, we have attempted to meet a number of new demands faced by traditional sensor-based measurement methodologies at a large scale, namely, a secure, sensor-based system that is easy to use for all users, while ensuring the rapid collection of accurate and useful data. In the process of producing the system implementation outline (Figure 4), we identified 3 key areas of risk.

First, we identified the risk of low recruitment and representative participation. Previously, sensor-based measurement on a large scale, whether it be for surveillance or cohort purposes, has struggled to recruit representative samples of sufficient size [23]. Through the early establishment and engagement of user groups, this risk can be mitigated, particularly by designing a system that overcomes barriers to use and fulfills user needs and desires. As this risk is context-dependent (eg, country, culture, institutions), future administrative users (groups 1 and 2) will need to consider how to ensure sufficient participation in the context where the system is implemented, and to adapt their approach accordingly.

Second, we identified the risks of remote sensor attachment, poor sensor utility, and poor adherence. Our system implementation outline relies on (1) the participant being able to attach the sensor easily, (2) the sensor being appropriate for free-living measurement, and (3) a high participant motivation to adhere to the procedure over several days. We plan to seek early input from the participant group (group 3) on the ease of comprehension of attachment instructions, and have repeated data collection throughout phases 1-3 to improve this aspect. Further, we will use phase 2 to provide an indication of the utility of the sensor in free-living settings and to maximize the usability of the system, which can encourage better adherence.

Third, with such a complex technological system, there will always be a risk of technological failure. The system implementation outline relies heavily on technical infrastructure for data transfer (from the sensor through the app to the cloud storage), automated analysis, and the provision of feedback. If any of these steps fail, we would risk losing data. We plan to integrate existing tried-and-tested technical infrastructure to mitigate this risk.

The strength of this protocol is the continuous inclusion of stakeholders throughout the development and evaluation of the system. Further, the development and evaluation of the system will be conducted in iterative cycles, allowing for continual improvement and incorporation of user input. A limitation of the current protocol is that the evaluation of the system will be limited to the social and cultural context that it is tested in. Each of the challenges and risks highlighted above will vary in degree depending on where in the world the system is tested and the culture of that place. Thus, the evaluation of whether the system is feasible or not will only be applicable in a Danish context until tested elsewhere. Future projects could consider including methods such as ecological momentary assessment to better understand the influence of context [24]. We have also opted to implement a single thigh-worn sensor. This placement has some limitations and the use of a single sensor does not allow for the capture of important physical activities such as awkward postures (ie, those deviating considerably from neutral positions)—for example, elevating arms above shoulder level and bending of back, kneeling, and squatting. Finally, in this step, we focus only on adults of working age, limiting our findings to this population.

The SurPASS system developed in this project will be just the first step toward achieving the SurPASS team’s vision. Further evaluation will undoubtedly be required to consider and improve on any transparency, accessibility, scalability, replicability, interoperability, privacy, security, and confidentiality issues that may arise as the system is adapted and challenged in new contexts (eg, low- to middle-income countries). In this first step, we consider the privacy, security, confidentiality, and transparency issues within the Danish context through the development of data usage and licensing agreements, as well as provide indications of future scalability through cost analysis. If the SurPASS system were feasible, it would represent a huge leap forward in the process of developing surveillance tools to meet the new demands on physical activity surveillance. The SurPASS system could facilitate the production of more accurate and useful data, which can be rapidly provided to key stakeholders once the validity of the sensor has been fully established. We believe that by developing a system that is low in burden, the prevalence of sensor-based measurement will increase, perhaps reaching population groups that are seriously underrepresented currently (eg, low- to middle-income countries, lower socioeconomic status settings). A higher prevalence of sensor-based measurements would also lead to increased public awareness regarding the importance of daily physical activity, sedentary behavior, and sleep (eg, through the rapid provision of accurate feedback). However, feedback to users is something that will need to be continually optimized as we gather better user information on how feedback is perceived to ensure it is relevant for users. Finally, many of the challenges encountered in the surveillance of physical activity are also faced by cohort and interventional studies, particularly burden and cost. The SurPASS system could be adapted for cohort and intervention research, further expanding the capacity of research programs and consortia [25,26] to provide high-quality evidence to inform policy and practice. The SurPASS project started in September 2020 and is currently in phase 2. The first results of the development and evaluation will be available in 2022, with the results of the feasibility evaluation made available in 2023.

The protocol presented describes a system that, if feasible, could act as a catalyst toward large-scale, sensor-based surveillance, thus taking the first step toward advancing physical activity, sedentary behavior, and sleep research and elevating evidence-informed policy and practice.