Keywords

1 Introduction

User-centered design (UCD) is vital for the design and implementation of technological and computer-based devices. Personas are a large part of UCD however there is limited research on describing the persona creation process in the health IT field [1]. Α persona is typically a fictional representation of the target group (or market) of the technology or device, which can effectively communicate to developers the requirements, wants, needs and motivations of the group [7, 9]. Research from Holden and colleagues [5] maintains that while personas are used extensively in many fields, there is a dearth of persona use in the health sector where e-health or health technology is concerned.

When creating a persona, high level information will often be included such as names, gender, age, location (e.g., country/living status etc.), photos and personal attributes. More key information of interest to the developers will support the development of their technology or service. According to some research [12], personas can be used by developers to support qualitative research sessions with a group of individuals. They can often be a discussion point, providing a way for group members to examine the life of the persona in a way that is less personal making them more inclined to discuss issues. As a result, researchers can more clearly identify the motivations and requirements of the user group.

In addition, recent trends in eCare and eHealth markets are showing patients evolving into consumers as their share of health and care spending grows continuously [3]. This trend opens opportunities for innovative technologies to increase competition and drive further disruption into the market. Consequently, stakeholders will gain the means to be informed decision makers. This opens the ability to integrate feedback from different sources (persona) already in the development process and to consider stakeholders’ interactions, values and unique selling propositions (USPs) in the design of business models and value-chains.

Personas can more than assist developers to construct and test their prototypes. The prototype will be developed with the persona in mind and during testing, participants can discuss how the persona may interact with the technology and their experiences of doing so. In 2019 research [6] considers how personas can also be used to make more effective decisions in the design of technology and services. From a HCI standpoint, the persona will influence developments from the earliest stages. Personas can also be used by organisations to demonstrate to target groups how technology or a service can be useful to themselves in their daily lives. Further, it can assist organisations in their marketing strategies. Research [8] was conducted to determine the specific benefits of personas on research based on the responses of experienced persona users. Their panel of experts identified several main benefits to persona use in research. These included re-focusing designers priorities on the user and their goals, prioritizing the requirements of the product/service and, ensuring designers do not design with their own requirements in mind. A lack of consistency was identified [5] in persona development and the lack of specific steps required for their development. However, personas can be developed using thematic analysis, affinity diagrams or even factor analysis of quantitative data. This view was supported [8] and further suggestions were made inferring that the development process is often incorrectly viewed as a singular procedure; something to be accepted or rejected, when in fact there are numerous methodologies which can be applied.

There are two major categories of personas, a proto-persona and a research persona [6]. Proto-personas are often developed as a precursor to work with participants. They involve the estimation of life details and are developed through indirect contact with the target group. They are based on assumptions and can often be influenced by stereotypes [8]. This assumed knowledge can enforce believability in a persona and ensures their ‘story’ is engaging [8]. While stereotypes can ease the cognitive load associated with the creation of personas [1], developers should be alert to the pitfalls associated with ethnicity and gender. While stereotypical personas may result in unintentional confirmation bias, they can also be used as an effective contradictory tool. Initially, the personas designed further validated the perspective of the frail older adult. However, the new personas better reflect a modern older adult perspective complete with varying abilities in digital literacy (a concept not previously included).

Further steps were recommended [6] when developing proto-persona and research personas. These include; creating a basic template, adding details, analyzing and merging personas, prioritizing important features and, design informed by qualitative data collection. The proto-persona will inform the development of the research persona. In this way, an enhanced persona is created which can lead to a ‘gap analysis’ to identify the breath of distance between the personas to provide developers with further insight in future persona development. However, some research [6, 9, 10, 13] has demonstrated serious faults with persona development. A time consuming process, the personas age rapidly and require regular updating and testing for validity. Often, personas are not used early in the design process and are less effective when design features have already been decided upon. Further faults were described in [9] including low representation rates when compared with big-data and the increased cost manual persona development incurs.

After completing this review, it is clear that the methodology described below is the best formulation to attain well developed personas based on the type of research being conducted. CAPTAIN is an iterative piece of design research with a small sample size and no real-time data capture abilities (at the time of writing). As such, initial proto-personas will be designed, followed by the creation of research personas. It is important to be able to validate the quality of the designed personas. Along with statistical demographics which will be discussed in the results section of this article, an evaluation of the personas based on the persona perception scale (PPS) [11] will be included. The scale distinguishes persona validity through the following dimensions, credibility, consistency, completeness, willingness, usefulness, empathy and similarity. Participants completed the PPS for both sets of personas (original and enhanced) and the results are discussed below. As a result, this paper is uniquely positioned to provide a review on the development of older adult personas in general, and specifically an account of the application of the PPS scale.

2 Methodology

2.1 The CAPTAIN System

The research presented in the current paper was conducted during the H2020 CAPTAIN project [4]. CAPTAIN aims to create an unobtrusive, virtual coach assistant that will support older adults in their everyday life at home. CAPTAIN provides support in four domains, cognitive and physical activity, nutrition and social participation using technologies for unobtrusive monitoring like 3D depth cameras for recognition of movements, speech recognition and generation as well as facial emotion recognition. The information is provided to the user through projection generation wherever and whenever needed leaving the home untouched when the system is not used.

Given the innovative nature of CAPTAIN, based on radically new ICT concepts an incremental, iterative delivery and empirical feedback approach was adopted that followed co-creation principles and methodologies was adopted. An active, trans-national and multidisciplinary community of stakeholders was engaged as the main source of requirements for the CAPTAIN system. This stakeholder’s community has been meeting throughout the project’s life cycle in order to assess the direction of the individual components of CAPTAIN during shorter development lifecycle, referred to as “Sprints”.

In order to engage stakeholders early enough in the design process, the first version of user requirements produced for the system was based on literature and consortium expertise. User personas were created to gain a deeper understanding of the users and these were used to enhance the methodology and capture user expectations and anticipated behavior beyond demographics. CAPTAIN personas had a dual contribution, building understanding and empathy in the consortium and in end-user comprehension of the system’s objectives.

2.2 First Matching Iteration

The first version of personas for the CAPTAIN system were created during face-to-face plenary project meeting. Five user personas were presented, that were based on the project’s objectives and user groups, demographics and ethnographic details. Discussion was held on ideas that can offer some solutions to the problems, thoughts, fears and opinions of the personas previously created. The proposed template for the CAPTAIN proto-personas is shown below Fig. 1. The proposed solutions were gathered and consolidated in order to create a structured version of Personas. After that, one round of feedback from the stakeholder community took place, involving 2 older adults (1 healthy, 1 with mild cognitive impairment), 2 facilitators and 2 formal caregivers.

Fig. 1.
figure 1

CAPTAIN persona template

Full size image

The resulting personas were used in participatory workshops in which older adults and other stakeholders were engaged, and the primary goal of which was to identify older adults’ everyday life problems and needs. A larger group of participants (52 older adults, 29 healthcare professionals, 10 informal caregivers) were engaged during the first co-creation session in 5 European countries (Greece, Italy, Spain, Ireland and Cyprus) and provided feedback that was incorporated into the personas.

Following the creation of proto-personas and the enhancement of same from user feedback gathered, the intention was to investigate if the group of personas created can effectively represent the group of stakeholders that participated in the session. For each participant an enrolment booklet requesting mainly demographic information was collected. The existing personas also contained such information. The following table (Table 1) summarizes the demographics information that are available for each participant and the relative information from personas.

Table 1. CAPTAIN participant general demographics
Full size table

A matching procedure was defined in order to understand how many participants can be adequately represented by existing personas. For each category in the previous table, actual participant’s information matching the persona’s information receives 1 point towards the persona. For example, participant GR001 is in the same age range category, living condition and has the same physical impairments as the persona ‘Carlo’, but does not share the same needs (2 needs per persona) so achieves a score of 3/5. If the score was 0.6 or higher, the participant was considered as matched with the persona. If a participant was matched with more than one persona, the higher score was taken into account.

2.3 Persona Enhancement

Following the first analysis of proto-personas, it was obvious that there was important information missing from the personas. First and foremost, information about digital literacy, technology acceptance and perception are an important addition to the personas profile. The digital literacy score was based on the aggregated score from a questionnaire that was administered to the participants before their first interaction with the system. Digital literacy refers to the participants level of comfort using technology and participants scores were grouped as low, medium or high.

In order to create the new personas, a prioritization of information that will drive the persona enhancement was completed. The real participants are considered to be representative of the CAPTAIN target audience. Age and gender were considered as important demographic information. ‘Living condition’ is a factor that can influence a user’s acceptance of the CAPTAIN system which is designed for in-home use. Low digital literacy may dissuade people from using a technology like CAPTAIN, which is why people with different digital literacy levels should be addressed, both in the personas and in the feedback Sprints. Lastly, the willingness to be supported and the impairments are considered the least significant factors that influence CAPTAIN system’s acceptance. As a result, the priority of information was decided as:

  1. 1.

    Gender

  2. 2.

    Age

  3. 3.

    Living condition

  4. 4.

    Digital literacy

  5. 5.

    Willingness to be supported

  6. 6.

    Impairments (motor and sensory)

For each enhanced persona, the gender and age were decided first based on the gender and age distribution of the real participants. For the remaining information a similar approach was followed, always taking into account real participants’ data. The information gathered for each persona were then analyzed in order to create a consistent story. Particular attention was paid to coherence and not to include multiple pieces of information that diverged from persona’s scope and personality. Last but not least, the personas’ photos were decided.

After the profiles of the new personas were created a new matching iteration was performed. Each new persona has a maximum score that can be achieved from a real participant as each has different types and numbers of support needed as well as deficiencies. A participant is considered matched to a persona only if the score percentage is higher than 60% and each participant is matched only with the persona having the highest percentage match.

3 Results

3.1 Persona Comparison

The enhanced persona creation was driven entirely by the participants’ information. The gender and age distribution are presented in the following image (Fig. 2).

Fig. 2.
figure 2

CAPTAIN participants breakdown by gender and age

Full size image

Based on that data three male and four female personas were created, two of the personas were in the first age range, three in the second and two in the third. In the previous version of personas there were three male and three female personas, four of which were in the second age range, one in the first and one in the last. Regarding the living condition, in the enhanced personas four are living with a partner at home, two are living alone and one is living with an informal caregiver. In the previous version, there was no persona living with an informal or formal caregiver. The digital literacy group that the persona belongs to was decided taking into account both the living condition and the age range. For example, a persona that lives home alone and is in the higher age range will have low digital literacy (Fig. 3).

Fig. 3.
figure 3

Distribution of participants ages and digital literacy versus living conditions

Full size image

Below we present an original (Fig. 4) and the corresponding enhanced (Fig. 5) persona for comparison.

Fig. 4.
figure 4

Original persona

Full size image
Fig. 5.
figure 5

Enhanced persona

Full size image

An important difference between the two personas is that the digital literacy that was included in the enhanced persona completes the narrations and improves the consistency. Some information is removed, e.g., “…going on road trips and driving his car”, as it does not add to the persona’s narrative or personality and seems like surplus information.

3.2 Persona Mapping

The objective of this work is to define personas that can adequately represent the actual users of the designed system. For this reason, two mapping iterations were performed, for the original (Fig. 6) and the enhanced (Fig. 7) persona version., following the methods described in the Methodology section.

Fig. 6.
figure 6

Distribution of real participants to original personas

Full size image
Fig. 7.
figure 7

Distribution of real participants to enhanced personas

Full size image

In the distribution of real participants across original personas, we notice that most participants are not matched with any persona and the user group is under represented. Though some participants may not be fully represented by the enhanced personas either, the distribution is wider.

3.3 Persona Perception Scale

Personas are a widely used technique in many technology-driven fields. However, until recently, there was no way to accurately assess the personas that were developed. The persona perception scale [11] was created to provide researchers with a way to test and validate the accuracy and credibility of development personas. The scale assesses personas based on a series of factors that are described in detail in their 2018 paper. The scale was assessed (Table 2) and high reliability scores were identified.

Table 2. The persona perception scale – factor reliability
Full size table

Upon analysis of the PPS, it was determined that there was no statistical difference between the original and enhanced personas. Inspection of the general descriptive statistics identified an interesting gender difference. PPS scores all improved for the male personas whereas, the female personas all got worse after being enhanced. The small sample size of our study may also have limited our ability to detect a significant difference (Fig. 8).

Fig. 8.
figure 8

Persona perception scale scores for original vs enhanced personas

Full size image

4 Discussion

As a part of the agile methodology of the CAPTAIN system development, older adults were invited to discuss the wants, needs and requirements to create a system that would support them in the home. To assist these discussions, six personas were created. These personas included some personal information including; age, gender, marital status and some physical/sensory/cognitive impairments the persona may experience. These personas were supplemented by images (See Fig. 9 below) to help create a more realistic character.

Fig. 9.
figure 9

Original Persona Images

Full size image

In the group sessions, only five personas were used due to time con-straints. Participants were asked to discuss the issues that may be experienced by these personas and what solutions may be offered to help these characters. Some participants would have preferred that the images used were a little younger, whereas others found the biographical information supplied to be an issue. Participants in Ireland found the personas to be unrealistic and at times, stereotypical. Participants suggested that the persona descriptions suggested that the personas were much ‘older adults’ than they represented themselves. Researchers also found some cultural differences across the personas which were a challenge. From a practical point of view, continental European health services appear to offer a substantial amount of support in the form of homecare assistance. However, in Ireland, participant’s joked that individuals would be doing well to receive two hours of similar assistance. As a result, Irish participants did not feel the personas were realistic. Further, participants in Ireland had a more ‘get-up-and-go’ mentality, with very little patience for personas who were unable to manage their health or who spent undue time worrying about it. There was an expectation that everyone should just ‘get on’ with things, regardless of the issue. As such, this was a barrier to the Irish participants identifying with the personas in the way the researchers had hoped.

As populations age, and with over 149 million adults over the age of 65 expected to reside in the EU by 2050 [2] and more research is con-ducted with these population participants’ concern about the stereotyping they felt was used in the creation of the personas is an issue of high importance. While some research [8] discusses stereotyping in persona creation, it focuses more on the gender misrepresentations rather than ageist views. Older adults are not satisfied with past views and many want others to accept that they are healthy and involved in their lives. The language and images used to create personas for adults over 65 is a vital area in the creation of realistic and relatable narratives. Terms like ‘oldest old’ to describe the older, older adults (85+) are used throughout the literature. However, even this term is being viewed negatively by participants.

After creating the new personas (see Fig. 10) based on the statistical demographics collected from 70+ older adults, they were offered to participants to evaluate based on the persona perception scale. The scale included items such as ‘the persona seems like a real person’ and ‘I feel like I understand this persona’. These questions provided a complete view of participants’ views of the personas. As such, this questionnaire was offered with the original personas along with the enhanced personas to provide researchers with information which could be directly compared. The results indicated that while there was no statistical significance between the original and enhanced personas, the male personas did receive more favorable results. This is interesting as only the male personas were improved upon, whereas the female persona scores got worse. Further study and discussion is required to determine whether there is a reason why participants found the female personas to be lacking. However, upon reflection, the researchers felt that simply asking participants to score a series of personas may not be enough for them to fully evaluate them. A simple group discussion before each personas is scored may have allowed participants to ‘get-to-know’ the persona first, before evaluating them.

Fig. 10.
figure 10

Enhanced persona images

Full size image

4.1 Limitations

A further issue with the development of the personas was identified just after the initial data collection. There was confusion surrounding the term ‘willingness to be supported’. Some of these issues arose due to language constraints and cognitive difficulties. Asking some older participants to imagine if ‘they would be willing to be supported in the future if they developed a series of difficulties’ was a challenging concept to understand and translate. After further discussion with participants another factor was identified which may have led to difficulty answering this item of the questionnaire. For younger older adults who were asked to report what areas they would be willing to be supported in, in the future, ‘the future’ was a further away concept than for the ‘older’, older adults. As a result, researchers thought it was possible that statistically, ‘older’ older adults who took part in the study would be less likely to agree to support in areas. However, there was no statistically significant difference between the two groups of older adults. Rather, with the exception of activities of daily living (ADL’s), where more ‘younger’ older adults were willing to be supported, ‘older’ older adults were either as likely or more likely to be willing to be supported in the future. There is a possibility that older adults may be under-reporting their willingness to be supported in group discussion settings. By answering the question on willingness, the participants may be faced with exposing potential vulnerabilities or acknowledging concerns they have for the future. This may result in a form of dissonance which prevents the participant from thinking about the question fully. Using alternative methods might elicit responses of more detail and fidelity.

The presented personas and persona development methodology depends entirely on data from older adults. However, the CAPTAIN stakeholders’ community and potential customers comprised a wider user group including healthcare professionals, informal and formal caregivers, and older adults’ family members. Data from these groups can also be exploited to create more personas. This work focuses only on older adults as they are the primary user group of CAPTAIN technology.

4.2 Future Directions

Unlike the proto-personas and the research personas discussed above, data-driven personas incorporate a continuous stream digital data collected while the user engages with a service or piece of technology [9]. Data-driven persona development also addresses the issue associated with research personas. The time spent developing personas is decreased significantly and personas can be updated regularly [10]. However, this process cannot be completed without any human interaction. Due to the current limitations of technology (i.e., natural language processing), and issues such as the perception of personas including persona bias, human intervention is required.

There are several methodologies which can be used to direct data-driven persona development but these all require large quantities of quantitative data to be collected while the users are engaged with a service or piece of technology. Due to the small number of participants included in the research and the nature of data collected, it was not possible to use this methodology to inform persona development. However, as this process is being enhanced step-by-step (proto-persona, research personas) it is possible that when the CAPTAIN technology is available the real-time data that is collected could inform a round of data-driven persona development. As technology continues to advance, the amount of personal data which can be collected from technology is increasing. Effective data management, the regulatory environment (e.g. EU-GDPR and the ethical implications of data-driven persona development must be considered.

5 Conclusions

User-driven innovation is a key competitive factor for CAPTAIN; in this regard, one essential focus point for the definition of the value-generation chains and the service delivery models is not only the identification of stakeholders, but moreover their involvement to understand, at an early development stage, their expectations and the degree of acceptance of proposed systems. Persona development and research is valuable to empathize with the stakeholders and improve system acceptability. Furthermore, personas are valuable tools that can be used for customer validation and market analysis. Personas are potential customers that provide information to the marketing team on how to empathize with them and create targeted marketing plans.

The presented personas and persona development methodology depends completely on data from older adults. However, the CAPTAIN stakeholders’ community and potential customers are comprised of a wider user group including healthcare professionals, informal and formal caregivers and older adults’ family members. Data from these groups can also be exploited to create more personas. This work focuses only on older adults as they are the primary user group of CAPTAIN technology.