1 Introduction

Recent evidence shows that the worldwide obesity rate is increasing and has more than doubled since 1980 [1]. The latest data from the Centres for Disease Control and Prevention (CDC) showed that more than one-third of US adults were obese, with adults aged 60 and over more likely to be obese than younger adults [2]. Similar trends have been reported in Europe, where between 30 and 80 % of adults are obese with higher prevalence of obesity among men than women [3]. Obesity and being overweight have considerable effects on morbidity and mortality through various diseases such as type 2 diabetes, cardiovascular diseases and metabolic syndrome [3]. Obesity yields negative economic consequences as a result from direct costs (increased medical costs to treat the associated diseases), indirect costs (lost productivity due to absenteeism or premature death) and intangible costs (psychological problems and poorer quality of life). Fry and Finley [4] revealed that the total direct and indirect costs for the 15 countries that were European Union members before May 2003 were estimated to be €32.8 billion per year, while for United States the estimated cost was US$147 billion [5]. Nagai et al. [6] reported that in spite of shorter life expectancy, individuals that are obese still require higher lifetime medical expenditures than those of normal weight. These costs are likely to increase as the prevalence of obesity increases [7].

The above factors have prompted initiatives to prevent obesity which range from the implementation of public health policies such as nutrition labelling [8], unhealthy food and drinks tax [9], to interventions by means of mobile phone applications to promote healthy eating and increased physical activity [10]. Some literature reviews, which have studied the effectiveness of obesity prevention through mobile phone applications, reported a mixed outcome of the intervention effectiveness [11, 12]. Although these reviews suggest possible factors that contributed to this phenomenon, none of them looked into the design process of these applications and how it might affect their successful adoption by the target users. It has been widely acknowledged through a variety of studies that failure in understanding users during design and development of any product or system can results in a low acceptance and effectiveness [13]. One approach to ensuring successful product or system design is the application of user-centred design (UCD), first introduced by Norman and Draper [14]. UCD refers to how end-users influence a design through their involvement in the design processes and has been shown to contribute to the acceptance and success of products [15]. This paper presents a review of the extent to which the design processes of existing mobile phone applications to prevent obesity have incorporated the principles of UCD. Our objectives were: (i) to identify key principles of UCD that were applied to develop existing mobile applications to obesity prevention; (ii) to analyse the strength and weakness of their design approaches and processes; and (iii) to identify any gaps in the research and propose future directions.

This paper begins by describing UCD in detail and is followed by explanation on how studies that were included in the review were identified and analysed. Next, the results of the review are explained in detail. The last two sections of the paper discuss the strengths and weaknesses of the reviewed studies and emerging issues and future research questions that need to be addressed to advance our understanding. This paper’s main contribution lies in the identification of gaps within this research area from a UCD viewpoint and how to address these gaps through recommendations for future research.

2 User-centred design

UCD is a common term, encompassing a philosophy and variety of methods, which refers to how end-users influence a design through their involvement in the design process. The level of user involvement in UCD varies and can range from a simple observation of end-users in their working environment to including user representatives on the design team. Key principles associated with UCD (ISO 9241-210:2010) are described as follows:

  • Clear understanding of users, tasks and environments. Explicit understanding of the characteristics of users, tasks and environments enables identification of the context in which a system will be used by users. This context of use subsequently assists in establishing users’ and/or organisations’ requirements and relevant usability goals. Approaches such as stakeholder identification and analysis, field study, user observation and task analysis can be adopted to gain an understanding of users, tasks and environments [16].

  • User involvement throughout design and development. Active user involvement should be upheld throughout the design and development process of a system. This could be achieved through various ways, such as including end-users or their representatives in a design team, consulting potential end-users and relevant stakeholders to assist requirements gathering and involving end-users in usability testing [17].

  • Driving and refining design through user-centred evaluation. This key principle emphasises the importance of user-centred evaluation to inform a design and to improve it at all stages. Typical activities include presenting low or high prototyping and storyboarding to potential end-users, post-experience interviews and satisfaction questionnaires of preliminaries design, etc. [18].

  • Iterative design process. Making the design process iterative is a way of ensuring that users can get involved in the design and that different kinds of knowledge and expertise can be brought into play as needed [15]. Iterative design processes can be identified through integration of the formative evaluation outcome into later or final designs.

  • Addressing the whole user experience. This key principle emphasises how a design should also consider the quality of a user’s experiences while interacting with a specific design and not focus solely on usability, i.e. whether or not a design is effective and efficient. In other words, a design should promote positive emotions and feelings to users while interacting with it [19]. Attempts to address user experience can be identified through the use of interviews and/or distribution of questionnaires which probe end-users’ experiences after using a system.

  • Inclusion of multidisciplinary skills and perspectives. A range of views, including those of non-technical specialist experts, end-users, relevant stakeholders, etc. is required during the design and development of a system [15]. This could take the form of a consultation with and/or inclusion of these people in a design team.

3 Methods

The articles included in this review were primarily identified from a meta-search on engineering databases which included ANTE, ACM DL, Ei Compendex, IEEE Wiley eBooks Library, IEEE/IET Electronic Library, INSPEC (Ovid), INSPEC Archive (Ovid) and Zetoc. The search was limited to English language communications in peer-reviewed journals, conferences and books that were published between 2000 and 2012. Only articles that reported the utilisation of mobile phones to promote healthy eating and physical activity were included in the review. The following terms “obesity phone”, “physical activity” and “phone” and “obesity”, “weight loss” and “phone” were used to perform the search. As this review focuses on obesity, articles that reported the use of mobile phone applications for other purposes, e.g. diabetes patient monitoring, rehabilitation, wellness monitoring for elderly, etc., were not included in the review. Furthermore, articles, which focused on sensing devices and their mechanisms, or were limited to blueprints/concepts of mobile phone applications to prevent obesity, were also excluded from the review.

For each study, we asked the following questions: (1) “How is the created mobile phone application used to prevent obesity?” (2) “What key principles of UCD were adopted during its design and development process and how were they applied?” (3) “Was the mobile phone application effective and what measures were used to determine its success?” For the second question, design and research activities that were relevant for each key criterion were identified. It is acknowledged that the outcome of the assessment was largely affected by how much details of the design and research activities were reported or published. To minimise the effects of this issue, all publications related to a website or mobile phone application were tracked and included in the review. For the third question, only studies that reported deployment of their applications to users in real-life situation were reviewed.

The review method adopted in this paper is largely based on research literature from scientific publications and does not survey the many commercial mobile phone applications that have been developed in recent years. One reason for this decision is that, unlike for research literature, there is no systematic way to get a comprehensive overview of commercial products. The other reason is that our goal was to provide an overview of the design and development process of mobile phone applications and this information is not likely to be easily accessible in the commercially sensitive private sector.

4 Results

Figure 1 provides a flow chart documenting the results of the study selection process which resulted in the inclusion of 52 mobile phone applications in this review. As the review was based only on scientific publications and excluded those that are available commercially, it has to be noted that the list of the mobile applications for obesity prevention is not necessarily exhaustive.

Fig. 1
figure 1

Search results

Full size image

An overview of the mobile applications included in the review is shown in “Appendix 1”. A quick glance on this suggested that more than half of the applications design (33 articles—63.5 %) did not incorporate theories or principles that could encourage behavioural changes, despite aiming to do so. Although incorporation of relevant theories into applications does not necessarily guarantee the success of applications in inducing changes of behaviour, there is a strong likelihood that such a system would likely be more capable in achieving this through design. For instance, in mobile snack, Khan et al. [63] incorporated the transportation theories [13, 112, 113] and precaution adoption process model [114] by providing an immersive animation-based narrative that depicted the game characters’ progressive life stages based on their eating behaviour. They also applied social cognitive theory [115] by allowing users to view their eating behaviour history and compare themselves with other users. The overview results also suggested that only a minority of the reviewed studies defined the specific end-users for whom their applications were targeted and instead opted for a broad definition of end-users, e.g. individuals with obesity or overweight issues, individuals with sedentary lifestyle. Having such a broad definition of users will potentially result in overlooking some users and limit the applications’ effectiveness. For instance, users of UbiFit, a tracking application that provided feedback on users’ physical activity level through a garden metaphor on its glanceable display, preferred to have metaphors for displays that suited their interests [97, 98]. A similar thing was also experienced by users of Neat-O, a game application that motivates its users to compete against each other based on their physical activity level and rewards them with choice of mini games. In this instance, users suggested adjustments on the type of mini games provided based on age groups as well as pop-up motivational statements appropriate for different gender [73].

This paper revealed that mobile phone applications to prevent obesity can be grouped into four categories. These were as follows: a tracking assistant (assisting a user in tracking and reviewing his/her behaviour), an entertainment tool (persuading a user to adopt an intended behaviour through game), an advisory assistant (advising a user to adopt an intended behaviour) and a tool to leverage social influence (providing a platform for a user to interact with his/her peers to encourage adoption of an intended behaviour). Figure 2 shows the proportion of mobile phone applications in each category. The lower proportion of mobile phone application’s role as advisory and enforcement of social influence is likely due to the recent availability of mobile social network platforms and GPS technologies. As such, smart phone applications in these categories are expected to grow in numbers in the future. It is also expected that social network platforms and GPS technologies will be important part of future applications of the first two categories.

Fig. 2
figure 2

Four categories of mobile phone application’s role to prevent obesity

Full size image

A further examination revealed that the majority of the reviewed studies produced fully functioning prototypes (see “Appendix 2” for a review of the mobile phone applications from a UCD perspective). Unfortunately, it also appears that these prototypes were not necessarily developed based on a thorough understanding of users or how and when they would use these applications. Only 30.8 % of the reviewed studies indicated any design activities that involved either targeted users or relevant stakeholders prior to the design of applications. Incidentally, although the majority of the reviewed studies (88.5 %) either performed or at least planned an evaluation study that involved end-users, only less than half of the reviewed studies (42 %) demonstrated that the application designs were refined through user-centred evaluation prior to the development of fully functioning prototypes. Furthermore, only 46 % of the reviewed studies indicated the existence of an iterative design process. All of the above findings indicate that user-centred design principles were not fully adopted by researchers while designing existing mobile phone applications. These findings also suggest that the design for the functions and features of the applications was driven in the majority by perceptions of the application designers on users’ needs and requirements. Coupled with the fact that these studies adopted a broad definition of end-users, there is a strong likelihood that the end product is a mismatch for the targeted end-users. Unfortunately, it was not possible to identify from the reviewed paper what might cause this phenomenon. Vredenburg et al. [18] emphasised the importance of user involvement to support understanding of user and iterative development throughout the design and development of a system. During the early stage, user involvement is valuable to provide support for design decisions as well as problem detections of a system; while at a later stage, it can help to verify the quality of a system. Failure to provide adequate user involvement means that changes at later stages are likely to occur and be more costly or difficult to be accommodated. For instance, Buttusi and Chitaro [64], who designed Monster and Gold game application, discovered only after user evaluation that their application could be repetitive if used several times due to lack of game setting varieties. Another example is Fukuoka et al. [70], who designed mPED that sent daily messages to suggest context-aware physical activity, overlooked the users’ need for personalisation and variation of encouraging messages and only discovered at a late stage some of their end-users considered found this to be source of disappointment.

Most of the reviewed studies (63.5 %) addressed the whole user-experiences while evaluating their functioning prototypes. This finding suggests that designers acknowledged the importance of assessing user experience, especially when long-term engagement is required. All of the reviewed studies conducted interviews to investigate user experience. Although interviews enable an in-depth understanding, they present a challenge for comparison studies. Using a set of established questionnaires, such as AttrakDiff [116] is expected to resolve this issue. Last but not least, it was found that most of the reviewed studies (73 %) indicated the inclusion of multidisciplinary skills and perspectives during their application design and development. This proportion is high as various aspects such as the presence of underlying behavioural change theories, consultation with experts/users and inclusion of experts/users in the design team, were used as indicators.

Table 1 shows the evaluation results of the reviewed studies, which produced fully functioning prototypes and which were deployed to users for more than one day. Out of the 52 studies, only 14 (27 %) conducted field studies with their targeted end-users, with deployment of the applications ranging from one day to three months. This finding suggests that, at the moment, most of the research effort was concentrated on the development of the applications and lack of attention was paid to investigate their effectiveness. It is possible that this phenomenon occurred due to the continuous and fast-paced technological development related to mobile phones, which consequently encourages more research to investigate the latest technology utilisation and leaves little time to study its effectiveness. For instance, based on the reviewed paper, the incorporation of a mobile phone’s camera to assist automatic recording of food intake was only initiated in 2008, and as the time progressed, the complexity of technology used also increased, i.e. from simply using a camera [45, 46, 91–93] to video camera [39] and a combination of a video camera and laser-generated grid [37, 38].

Table 1 An overview of the evaluation of mobile phone applications included in the review
Full size table

The review showed that nearly all of the studies collected qualitative data (related to usability and/or user-experiences) and quantitative data simultaneously. The type of quantitative measurements that were collected by most of the studies differed from one study to another although all of them aimed to capture how their applications prompted behavioural changes. As a result, performing comparison of effectiveness between applications will likely be difficult if not impossible. Furthermore, taking into account that most of the applications were aimed at preventing obesity, it was quite surprising that only a few of the reviewed studies used weight change as one of their quantitative evaluation measurements. It is acknowledged that, as some of the studies only focused on one particular thing (either physical activity or eating behaviour), changes in weight do not provide an independent measure as it is affected by both physical activity and eating behaviour. With regards to the results of the evaluation, it is clear that more research is needed to provide firm evidence on the effectiveness of mobile phone applications to prevent obesity. All of the 14 reviewed studies suggested that mobile phone applications have the potential to encourage change of behaviour to prevent obesity. However, it is difficult to ascertain if the observed potential will be sustained on a long-term basis since only two studies out of 14 studies deployed their applications for 3 months. A similar problem has been experienced in establishing the effectiveness of internet based and video game to combat obesity [117–119]. Mattila et al. [108] even suggested that mobile phone applications will not be effective unless they are part of an intervention which “provides the initial motivation and engagement that enable users to reach the ‘long-term’ stage”.

5 Strengths and weaknesses of existing studies from UCD perspective

Based on the results of the review in Sect. 3, the strengths and weaknesses of existing studies from a UCD perspective were identified (see Table 2). Current studies have, to some extent, included multidisciplinary skills and perspectives through consulting and/or involving users early in the design and/or evaluation stage. However, there are limitations to the involvement of users as users’ knowledge of behavioural change domain and awareness of their own needs may be limited [120, 121]. Therefore, more effort should be made to incorporate relevant behavioural change theories into the design and development of mobile phone applications. A similar analogy to this is a creation of a virtual training system in which the training system’s content and delivery should still be based on training principles while its interface design and context of use are consulted with end-users or relevant stakeholders. As mentioned earlier, there is some evidence that users were involved during the development of mobile phone applications. However, most of the involvement occurred only in the formative evaluation with few, if any, taking place prior to this. Frequent end-user involvement could be impractical and costly as could perform major changes post-formative evaluation. Therefore, a design team should weigh the benefit and cost of each circumstance at the beginning of the project and use this judgement to make their decisions. The review also showed that iterative design practices were widely adopted across studies. However, the level of iterative design process varied between studies, with some studies only using the results of formative evaluation to improve their application design. Again, the same judgement made for user involvement explained above also applies. It should be noted that iterative design practices do not necessarily equate to higher level of user involvement in the design as the iteration could simply be triggered through less formal evaluation such as heuristic evaluation [122] or cognitive walk-through [123].

Table 2 Strengths and weaknesses of existing studies from UCD perspective
Full size table

Some weaknesses of existing studies were also identified through the review. The most commonly noted is the lack of focus on specifying target end-users. Obtaining a full understanding of users will likely be a challenge if end-users are not fully specified from the beginning. This could result in the inability to thoroughly identify what they want from a design, how and when they will use it, what makes them want to use it, etc. There is strong evidence that behavioural change needs to be tailored to match an individual’s needs and characteristics in order to be effective and sustainable [124, 125]. Furthermore, by providing specific target end-users from the beginning, relevant stakeholders could be identified and potentially involved. A perfect example of this is a study by Toscos et al. [31], who having specified their targeted end-users as teenage girls, identified and consulted relevant stakeholders (dieticians) who then provided valuable suggestions to their initial design concept. Another common weakness that was observed from the reviewed studies is the way mobile phones applications were evaluated for their effectiveness in inducing change of behaviour. The first issue is the length of period, which is considerably short to observe permanent behaviour change. The reviewed studies showed that the longest observation period was 3 months minus the first week to establish a baseline. This is considerably shorter than Prochaska and Velicer [126] recommended minimum of 6 months. It was also found that different studies utilised different approaches to evaluate their application effectiveness, which in turn created difficulties in comparing the studies’ results. A possible option to resolve this was to adopt a randomized controlled trial (RCT), which is commonly used in the health sciences. Fukuoka et al. [71] have published their detailed plan to evaluate the effectiveness of their mobile phone application to increase physical activity in sedentary woman. However, it should be noted that RCT may not provide an indication of why or how the mobile phone application could induce behavioural change. Therefore, if the aim of the evaluation is to find out the reasoning behind an induced behavioural change, additional qualitative data should be collected. There was also a question raised regarding the type of variables that were used to gauge the effectiveness of applications. For example, of two mobile phone applications that aimed to educate about healthy eating through game, one measured their application’s effectiveness based on the frequency of use [73]; the other based it on the frequency of users’ negative behaviour [95].

6 Emerging issues and future research directions

Following the increased acceptance of social network platforms and enthusiasm to share information publicly, it is expected that more and more mobile phone applications will enable sharing between peers. Sharing has been shown to have a positive relationship with inducing change of behaviour as a result of social pressure, support and accountability among members of a group [127, 128]. However, some of the studies that have incorporated this feature into their applications have reported issues that resulted from sharing such as unhealthy competitiveness [129], negative impacts if a desired response from peer was not received [30] and breach of privacy. More research will be required to avoid these issues while maximising the benefit of sharing between peers. Breach of privacy could also be an issue for mobile phone applications that integrate automatic sensing in various forms such as GPS, microphone and accelerometer. Klasnja et al. [130] reported that users’ perception towards breach of privacy is influenced by the type of data that is recorded, the context in which participants worked and lived and the value they perceived would be gained in return. They suggested preventive measures such as adopting conservative recording and data retention policies, graded functionality for the sensors, and giving users visibility and control over sensors usage. An issue that is unique to mobile phone applications with an advisory function is ensuring the appropriateness of suggestion with users’ lifestyles and activities as well as the ability to ignore it. Fukuoka et al. [70], who sent daily random message to their application’s users and required a subsequent response, found that only 39 and 34 % of users would like to receive it once and twice a day, respectively. However, a study by Mutsuddi and Connelly [131] showed that tailoring suggestions is less crucial when an application is deployed to users who are in the contemplation and preparation stages of change as they would likely more receptive to suggestion.

Throughout the review process of this paper, it was observed that there are limited guidelines on how to design effective mobile phone application interventions to sustain behavioural change. Although some research has begun to contribute to this topic, more research will be required in the future. Fogg [132], through his persuasive design concept, has proposed higher level abstraction that can be applied to persuade users while Consolvo et al. [100, 129] proposed several guidelines, which were drawn from the implementation of their applications and incorporation of behavioural change theories. Their proposal included appropriate reward for positive reinforcement, supporting social influence, taking into account the practical constraints of users’ lifestyles, reflection through abstraction, unobtrusive data collection and presentation, aesthetics, permitting the user to control their own data and enabling history/trend viewing. Another research direction for the future is a more comprehensive approach towards all dimensions of an individual’s life. Lenert et al. [133] suggested that the lack of comprehensive approach might be an avenue to engage users in the long term after an initial period of interests. A comprehensive approach is only feasible when information regarding users is obtained and understood thoroughly. This is crucial for mobile phone applications that aim to be context-aware. More research that is also needed in the future is establishing the effectiveness of mobile phone applications, which has already been discussed in detailed in the previous section.

7 Summary

As discussed above, from user-centred design perspective, the design and development process of mobile phone applications can certainly be improved by addressing its current weaknesses. Two main areas offered opportunities for improvement. The first is design of an application should be based on a combination of thorough understanding of users and their contexts of use, combined with principles from behavioural change theories. The second is a more robust approach in evaluating the effectiveness of mobile phone applications in terms of chosen quantitative measures, length of observation periods and how and why certain features are successful in supporting behaviour change. More work is required to establish guidelines that can be used to design and develop mobile phone applications to prevent obesity.