The goal of user-centered design (UCD) is to understand the users’ perspective and to use that information to shape more effective solutions [1]. UCD gains access to users’ needs and requirements and thereby improves the solutions’ design and increases their functionality, usability, and quality [2,3]. UCD can reduce the users’ need for support, decrease development costs, and increase user satisfaction and safety [4]. However, involving users in the development process does not guarantee that feedback from different subgroups of users will equally shape the development in such a way that the developed solution becomes useful for the entire user population for which it was intended. This study describes a method that enables systematic tracking of feedback from user subgroups in usability testing of digital support for fall-preventive physical activity (PA) in seniors (adults aged 65 years and older). The aim of the method is to support the development of a digital support system that is useful for different subgroups of users.

Digital support systems—for example, mobile apps or web-based services—have the potential to strengthen and complement existing health care resources. There is evidence indicating that digital support can be effective in somatic care [5]. The internet is also an important source for acquiring disease-specific knowledge in chronic care management [6]. In 2019, approximately one-fifth of the population in the European Union was 65 years or older, and more than 5% of the population was 80 years and older [7]. The proportion of individuals aged 80 years and older is expected to more than double within the next 80 years. Furthermore, approximately 80% of the population aged 80 years and older suffers from at least one chronic condition [8]. Provision of digital support to improve health in the growing population of seniors is therefore increasingly important.

It is well established that PA has a positive impact on health and well-being. For example, physically active people have lower rates of lifestyle-related diseases (including coronary diseases, high blood pressure, type 2 diabetes, colon and breast cancer, and depression) and exhibit higher levels of functional health and better cognitive function [9]. Despite this knowledge, physical inactivity is an increasing global health burden [9].

User involvement in the development of digital support systems is an evolving area. A review on development of fall-detection systems concluded that seniors are never involved throughout the whole development process and seldom involved in the early and end stages of development [10]. The authors emphasize that early user involvement, focusing on the users’ needs and preferences, is important for improving the seniors’ level of technology acceptance. There is a great variation regarding which users are involved in UCD and how they are involved. Moreover, reporting on the procedures used for the selection and recruitment of users is often lacking [11].

Seniors’ interest in and use of the internet is associated with sociodemographic factors including gender, age, and socioeconomic status [12-15]. Moreover, seniors use the internet for different purposes: while some use it solely for practical things, such as financial purposes, searching for information, and emailing, others use it for additional purposes, such as gaming and social interaction [16]. In addition, seniors have different needs and preferences for digital health care support, partly driven by their own experiences with health care [13]. The described diversity in the older population challenges the UCD because involving certain users does not entail that their feedback will represent views of the entire target population. There is a need to increase the understanding of user subgroups’ needs and contribution in UCD processes in order to involve representatives of relevant stakeholders in the development.

The aim of this study was to describe a protocol for systematic analysis and prioritization of feedback from user subgroups in the usability testing of digital support systems to motivate fall-preventive PA in seniors. This protocol can help researchers and developers to systematically exploit feedback from relevant subgroups in UCD.

This project is based on a UCD model [1] and the key principles described by Gulliksen et al [17], including early user involvement. An overview of the UCD process is presented in Figure 1. The protocol described in this article is used in study 4 (Figure 1).

Three key user characteristics have been identified in the literature to influence seniors’ experience of fall prevention, PA, and technology use: (1) gender, (2) PA level, and (3) level of technology experience. Gender was selected because a gender-based difference exists in PA behavior, falls incidence, and consequences from falls; men and women have different fall risk factors, both short- and long-term [18]. Compared to men, women are more prone to falling, sustain more fall-related injuries [19,20], and report fear of falling more frequently [21,22]. Men and women also differ in their exercise habits and reasons for exercise [23]: women are less likely than men to be regularly physically active [24] and tend to prefer different types of PA [25]. Gender has also been identified as crucial for understanding technology use [26]. PA level and level of technology experience were also selected because seniors with lower PA levels and less experience with technology might represent less engaged users of digital support services for PA. The 3 key user characteristics are integrated in the collection and analysis of user feedback.

By dichotomizing the 3 key user characteristics, 8 (ie, 2³) possible user subgroups were defined. The participant’s PA level and level of technology experience were classified as “higher” or “lower,” respectively, according to the user’s self-reported PA level and level of technology experience in the questionnaire on user characteristics (Multimedia Appendix 1). Participants who reported spending at least 3 hours/week performing moderate-intensity activities during a regular week (questions 1 and 2, respectively, on PA level in Multimedia Appendix 1) were classified as having a higher PA level, while participants who reported spending less than 3 hours/week performing moderate-intensity activities were classified as having a lower PA level. The threshold was chosen based on the PA guidelines for the adult population, which recommend at least 150 minutes/week of moderate-intensity aerobic PA [9]. The cutoff for higher level of technology experience was set to using a mobile phone and/or computer/tablet often for purposes including calls, texting, emails, and surfing the internet, as reported by participants (questions 3 and 4 on level of technology experience in Multimedia Appendix 1). This cutoff was chosen to ensure that participants had experience in surfing the internet. The participant’s gender was classified as man or woman according to the participant’s self-reported gender in the questionnaire (Multimedia Appendix 1).

Examples of subgroups were men with a lower PA level and higher level of technology experience and women with a higher PA level and lower level of technology experience.

The assignment of each participant to 1 of 8 subgroups was identified by analysis of the questionnaire on participant characteristics (Multimedia Appendix 1). Each participant’s user subgroup was visualized on all templates for processing and analysis of data generated by the specific participant.

During the iterative testing (Figure 1, study 4), the participants tested and evaluated prototype versions of the digital support (2 applications) in 4 test cycles. Each participant performed the tests individually with a researcher. Qualitative and quantitative data on use and use experience were collected during and after the test sessions.

Table 1 presents an overview of the methods for data collection, processing, and analysis, and concretization of improvements applied in each test cycle. Steps 1 to 3 were included in the method for systematic tracking of user subgroups’ feedback and they are described in detail below.

The method presented in this study provides a structured way to document and exploit feedback from different user subgroups in iterative development of new solutions, exemplified here by digital support systems for fall-preventive PA interventions in seniors. Involving seniors in technology development has been suggested as an important component for improving technology acceptance [10]. Moreover, physical inactivity is an increasing global health burden, and it is well documented that older age groups are less active than younger individuals [24] and that PA is important for improving health [9]. Moreover, although inclusiveness is an important design goal, discrimination is a common deficiency of digital support systems [14]. For example, gender inclusiveness and equality are important in system design, since the two aspects influence users’ behaviors, both online and offline [15,26]. GenderMag [30] is a systematic method for illustrating and tracking gender differences in software development by the use of personas. The method has proven effective for finding and fixing gender-inclusiveness deficiencies in software applications [31] and has been reported to be appreciated in practice [32]. Research on how to prevent discrimination and strengthen inclusiveness of digital support systems needs further attention.

In addition to increasing gender inclusion, the method presented in this study aims at ensuring that the feedback of users with varied self-reported technology experience and PA levels is shaping the development of digital support systems. The aim of integrating these perspectives is to encourage participation of persons who, because of lower levels of technology experience or PA, might have low interest in participating in UCD studies of technical solutions supporting PA. However, other user characteristics might also be of relevance when involving seniors in UCD studies. For example, Vandekerckhoven et al [11] suggested that creativity and communication are relevant and important user characteristics to consider in UCD studies.

In this study, the users’ gender, PA level, and level of technology experience were identified as relevant key user characteristics to define the subgroups to track in UCD studies. It could be argued whether these are the most critical key user characteristics to consider and if additional subgroups are to be included in order to support the development of digital support systems that fits the whole heterogenous target population. For example, fall history might be relevant to consider, since previous falls have been identified as a risk factor for new falls in both men and women [18] and it is vital to promote fall-preventive PA interventions in persons with increased fall risk. Moreover, users’ level of technology experience and internet use are related to age [16,33]: younger seniors (mean age less than 70 years) use the internet the most and older seniors (mean age 73 years) use the internet the least [16]. Furthermore, older seniors have more problems with activities in daily living and experience worse health than younger seniors [16]. However, this study used internet experience instead of age as a key user characteristic. To ensure that both younger and older seniors are involved in the UCD process, seniors representing different ages should be recruited.

Other examples of user characteristics that might be relevant for defining user subgroups include level of education and socioeconomic status [12,33], as well as personal qualities such as the level of communication and creativity [11]. However, in order to make the method feasible, the number of subgroups considered needs to be limited. This is accomplished by selecting user characteristics that are critical for the aim of the technical solution and can be dichotomized. In this study, 3 key user characteristics were selected based on the literature, and by dichotomizing them, 8 (ie, 2³) possible user subgroups were defined. The method represents a pragmatic approach, which is needed in UCD for balancing the quantitative research paradigm (requiring involvement of large user groups and thereby not feasible in UCD) and the qualitative research paradigm (focusing on one specific user subgroup and thereby not useful for an entire heterogenous population). However, further research is needed to validate if the key characteristics selected in this study are purposeful for creating a solution that is attractive and useful for the intended users of digital support systems for fall-preventive PA interventions for seniors.

User involvement alone does not guarantee that feedback from different user subgroups is correctly shaping the development of digital support systems and resulting in a solution that is useful for the whole intended user population. Further attention is needed on methods for systematic tracking of user subgroups’ feedback to ensure that new systems and services are designed for the entire target user population. Also, new knowledge is needed on how to select users to be involved in UCD processes. The method presented in this study elucidates and documents potential differences between how different user subgroups contribute to the development. This may clarify whether new users need to be added to the process, either to increase the contribution of specific user subgroups or to involve new subgroups.

Our hope is that this protocol will be used for systematic analysis and prioritization of user subgroups’ feedback in the development of new digital systems. The protocol may help to identify and improve the understanding of potential differences between subgroups of seniors in use and experiences of digital support for fall-preventive PA interventions. This new knowledge can be of great importance in future research to develop systems that are relevant, useful, and attractive to the heterogenous population. Moreover, it can facilitate tailoring solutions toward specific user subgroups. However, the protocol needs to be further used and evaluated to validate the potential value of the method and the purposefulness of the selected key user characteristics.