International Journal of Interactive Mobile Technologies (iJIM) – eISSN: 1865-7923 – Vol 17 No 02 (2023) Paper—The Development of a Fuzzy Model and Usability Test of a Recommended Interface Design for… The Development of a Fuzzy Model and Usability Test of a Recommended Interface Design for Mobile Phones for Elderly Users https://doi.org/10.3991/ijim.v17i02.33877 Weerapong Polnigongit1(), Waiwit Chanwimalueng1, Sandey Fitzgerald2 1 Institute of Social Technology, Suranaree University of Technology, Nakhon Ratchasima, Thailand 2 Macquarie School of Social Sciences, Macquarie University, Sydney, Australia weerap@sut.ac.th Abstract—The purpose of this study was to examine a range of flexible user interface designs for mobile phones with the aim of locating and addressing the limitations reported by elderly users. Accordingly, a fuzzy model drawing on a range of variables was developed from an age/vision impaired related data set for the development of a variety of basic design elements for user interfaces. The model was tested to assess the preciseness and accuracy of its functions, achiev- ing a Mean Absolute Error (MAE) close to 0 and an Effectiveness Index (EI) close to 1, giving the model a high value for effectiveness. A subsequent usability test of the generated design interfaces using four types of mobile phones (18 screens in all) was conducted among 25 elderly users with vision impairment. The findings showed that the size and shape of both numeric and function buttons was a significant factor in assessing phone usability both for communication and for social media use, as was text and number size, although, significantly, the latter was qualified by screen size. Recommended numeric and dial function but- ton sizes are 15.6mm and 16.2 mm, text and numbers sizes are 14 and 25 points, respectively. Square-shaped buttons with rounded-edge buttons are the most suit- able for elderly users, as is a background in a light shade, with texts and icons in dark colors. The model demonstrates that it is possible to design user interfaces with particular groups in mind such as the elderly and vision-impaired, in order to enhance mobile phone usability for these groups. Keywords—fuzzy model development, user interface design, mobile phone ap- plication, elderly mobile phone users, usability 1 Introduction Society is aging worldwide. The world’s population of elderly people more than doubled between 1980 and 2017, and is forecast to duplicate by 2050 [1]. These valu- able people face many challenges in modern daily life. One lies in the use of technology in the face of visual impairment. Although specialized training courses, especially com- puter-based training, may help find ways to compensate for visual field loss, loss of 118 http://www.i-jim.org https://doi.org/10.3991/ijim.v17i02.33877 mailto:weerap@sut.ac.th Paper—The Development of a Fuzzy Model and Usability Test of a Recommended Interface Design for… vision can never be recovered perfectly [2]. Elderly users, particularly those with visual impairment, will continue to have specific difficulties in using technology. This applies especially to what is currently the most economically accessible device, the mobile phone [3]. According to a 2011 survey of the use of technology devices by Walker and Masnard [4] involving 2,947 elderly participants in 23 countries, 87% of the partici- pants used mobile phones. In Thailand, a full-fledged ‘ageing society’ in 2021, the Na- tional Statistical Office [5] revealed that over 80% of elderly Thai people used mobile phones for communication. Moreover, among all applications available, 89.8% of Thai elderly people also spent their time on their phones on social media. Within this figure, 91.5% of elderly users chose to use LINE [6], a free application for smartphones as well as tablets and PCs, for instant communication by voice, video, image or text. However, a problem faced by these elderly users is that user interface designs (UIDs) for these devices are intended to cover all age groups of users [7]. They therefore may not meet the needs of specific groups of users such as the elderly and/or vision impaired. While users are expected to be able to adjust aspects such as the size of the operation buttons and typefaces of a device’s operating system (OS) themselves [8], it may not be easy or even possible for elderly users to manually adjust the complex settings on their mobile phones [9], especially as one of the most significant problems reported in relation to the components of mobile phones for elderly users relates to the phone’s touch screen [10], including ‘buttons,’ text and menus [11]. This problem includes el- ements of design such as size, shape, and color [11,12]. This study draws on Artificial Intelligence in order to come up with a user interface design for mobile phones that has the flexibility to meet the physical and cognitive requirements of elderly users, particularly those with visual limitations, in order to im- prove their experience of using a mobile phone device. A fuzzy model was developed to identify feature selection and create fuzzy rules for designing appropriate user inter- faces on mobile applications and social media applications by focusing on components of application and design that would enable the elderly to use an interface at its highest level of efficiency and effectiveness while meeting their satisfaction, the three aspects of the ISO 9241-11 usability standard. The model scored highly when tested for these three usability indicators, allowing recommendations to be made to application devel- opers. 2 Literature Artificial Intelligence (AI) is the attempt to design machines, particularly computer systems with processors and memory, which can simulate human intelligence, act ra- tionally and autonomously, and can learn [13]. A strong AI would be a computerized system that had ‘a mind.’ While this is proving challenging for computer scientists to develop, their efforts to develop such systems have already been beneficial in many fields involving complex decision-making such as medical science, economics, geol- ogy [14], general technology development, and the environmental sciences, where AI techniques enable modelling by drawing on a range of different methods, often in com- iJIM ‒ Vol. 17, No. 02, 2023 119 Paper—The Development of a Fuzzy Model and Usability Test of a Recommended Interface Design for… bination, including case-based reasoning, rule-based problem-solving involving ma- chine learning, the creation of artificial neural networks to mimic the way human brains operate and ‘genetic’ algorithms to mimic natural selection, ‘multi-agent’ systems in which components are placed into networks within which they interact, and fuzzy set theory [15]. AI technology utilizes algorithms or precise sets of instructions to respond to human instruction. Fuzzy logic, a development of fuzzy set theory first demonstrated by L.A. Zadeh in 1965, is one of the techniques used in AI studies. Unlike sets in classic theory (so-called crisp sets), which are binary, fuzzy set members are ‘fuzzy,’ like human thinking, with indeterminate boundaries because of uncertainty about where boundaries exist. Fuzzy sets approach the measurement of things by ‘scaling’ [16]. They therefore provide ‘a mathematical tool to deal with uncertainties’ [17]. Models using fuzzy logic can take into consideration conditions that are ‘vague or not precisely defined’ [18], allowing the solving of problems where vagueness of information emerges [19]. Fuzzy logic is therefore ideal for modelling decision-making or solving problems where un- certainty occurs [20]. It allows ‘the study of vagueness:’ whenever uncertainty emerges this theory allows models to be ‘constructed to represent and process’ specified prob- lems [21]. This ability to handle vague or imprecise information makes fuzzy systems one of AI’s ‘strongest techniques’ [15]. Developments in mobile technologies such as phones and other portable internet connectable devices in conjunction with developments in AI have increased the ‘appli- cation scenarios’ for AI enormously [22]. Scholars are already envisaging future smart 6G networks that will make the Internet of Everything (IoE) a reality for mobile device users [23]. AI assisted mobile phones already allow wallet-less shopping, convenient vaccination/health status display, smart home and health care management [22]. AI, including fuzzy logic, is being used to develop applications and mobile devices that are useful for those with visual impairment, including the elderly. Harum et al., for in- stance, have used AI technology and smartphone application technology such as the Digital Daisy Book Reader to develop a multi-language interactive book reading device that can be used by the visually impaired to ‘read’ information in public places [24]. Conversely, mobile apps for smartphones are increasingly being developed to help us- ers manage a huge range of human needs, from finding ways for dysarthic children (children with a neurological disorder that damages their ability to speak) to communi- cate using ‘daily usable conversation terms’ [25], to encouraging inveterate online tex- ters to find more polite ways of communicating [26]. Mobile phones are being devel- oped as ‘fall detection system’ sensors ‘trained’ to detect ‘falling in any direction’ from common activities such as walking or jogging [27]. Bratić et al. have also used a fuzzy logic-based model to evaluate the readability of handwritten font sizes on small devices such as mobile phones, finding that some font types in some sizes are more readable than others [16]. Clearly, AI is already widely used in smartphone technology, and likely to become more so as phones learn to automatically carry out tasks that are im- portant to users [28]. Applying AI techniques can conceivably generate a more-person- alized and adaptive service to users, if AI designers incorporate the needs of user, along with product development, principles, and processes [29]. Nevertheless, regardless of 120 http://www.i-jim.org Paper—The Development of a Fuzzy Model and Usability Test of a Recommended Interface Design for… the sophistication of the application, older mobile phone users generally choose a mo- bile phone for its apparent usability, based primarily on its functional attributes [30]. Unless the targeted customers for these wonderful services find the basic functions of mobile phones easy, comfortable, and satisfying to use, these applications will fail to be embraced. Therefore, it is important to understand the basics of what elderly people require to make mobile phones easier and more satisfying to use. AI is therefore in- creasingly being applied to test the usability of mobile devices. This is becoming even more important since it has been recognized that high levels of human-machine inte- gration can in effect reduce user autonomy, limiting the appeal and usability of some designs for vulnerable user groups such as the elderly [29]. However, although studies on the usability of particular applications for mobile phones have burgeoned, basic function usability (the hardware of devices) has not kept pace with developments in smart-phone technology, or with the needs of aging popu- lations. In 2005, Ziefle and Bay reported that although older adults were very interested in and willing to use new devices, they found the usability of the designs was not ade- quate to their needs or abilities [31]. In 2014, Kamel Boulos et al. argued for the devel- opment of application operability standards and certifications for mobile phones, since apps that were ‘perfectly usable by a younger person’ could be unmanageable for older or disabled persons with different usability needs, and increasing age combined with increasing complexity of interfaces in mobile phones were factors that resulted in poorer performance [32]. Yet, in 2020, Jiang et al., still found that current smartphone user interfaces, specifically those for camera use, were not optimized for users over 50 years of age, although this age group was a much greater user of mobile phone cameras than younger phone users. Many interviewees thought there were too many functions, and routinely relied only on a few, even when aware of others [33]. Huang’s 2020 ‘state-of-the-art review’ also found that application designs for mobile phones were much more constrained by the physical limits of the devices than for PCs and larger devices, but designers were not factoring this into their application designs. Nor were usability studies keeping up with later mobile phone actions such as swiping and pinch- ing touchscreens [34]. In general, although existing guidelines and checklists for de- signers had increased in complexity as mobile phones shifted from feature phones to smartphones and increased in complexity, there were many usability dimensions in re- lation to age-friendly design that were still not well covered. Factors affecting usability for older people that had been identified by scholars were also still not being picked up by existing guidelines, while the guidelines themselves were becoming so complex in their efforts to cover many more dimensions and categories that they risked being un- usable. Nor was there recognition that age-related guidelines were not static: the tech- nological skills of older users were increasing over time as middle-aged users aged, even as the usual physical limitations of age (memory loss, failing eyesight, loss of dexterity and hand control) remained much the same [35]. To date, design technologies have tended to treat the elderly, including those with visual impairment, as ‘passive receivers’ [36], with limited recognition of their specific needs. iJIM ‒ Vol. 17, No. 02, 2023 121 Paper—The Development of a Fuzzy Model and Usability Test of a Recommended Interface Design for… 3 Research hypothesis The purpose of this study was to generate and test a model flexible user interface design application for mobile phones with the aim of addressing usability limitations reported by elderly users of mobile phones. In developing a fuzzy model for this pur- pose, the hypotheses were: 1. [W]hen the value of the Mean Absolute Error (MAE) of the fuzzy model for de- signing user interface applications for mobile phones gets close to 0 and the Effec- tiveness Index (EI) gets close to 1, the three aspects of usability according to ISO 9241-11 (efficiency, effectiveness, and user satisfaction) for a user interface appli- cation on a mobile phone would be at a high level. 2. User-testing this model for efficiency, effectiveness and user satisfaction would pro- vide viable usability recommendations of a high level to be made to application de- velopers. 4 Methodology Most user interface designs are designed to meet the general needs of all ages of users. This results in usage problems and dissatisfaction for specific groups, such as elderly users [7]. Therefore, the usability of such interface designs should be evaluated. Although Nielsen [37], Sharp et al. [38], and Quesenbery [39] explain usability from different perspectives and it seems that usability can be tested in different ways, there is general agreement over five key principles of usability: task efficiency; easy to learn; rate of error; memorability; and satisfaction. The International Organization for Stand- ardization has reduced these to three key principles in their ISO 9241-11 usability standard: efficiency, effectiveness, and satisfaction [40, 41]. These are the principles this study utilizes to measure the usability of the recommended interface designs. The research procedure for the study was divided into two parts. The first entailed the de- velopment of the fuzzy model. The second entailed the usability evaluation of the model. 4.1 Research tools The research tools for the study covered three areas: 1. Research instruments for the data collection procedure: Data collection was imple- mented using data collection computer software as a research tool, drawing on in- formation supplied by surveying/interviewing a preliminary group of participants. This included their responses to a list of 40 questions divided into 15 multiple-choice questions (demographic and human factors) and 25 estimation-scale questions (de- signed for self-adjustment). 2. Research instruments for the fuzzy model development process: The tools included in the process of developing the fuzzy model included the Waikato Environment for 122 http://www.i-jim.org Paper—The Development of a Fuzzy Model and Usability Test of a Recommended Interface Design for… Knowledge Analysis: WEKA version 3.6, the Matrix Laboratory: MATLAB version R2014a, and the Development Application for Data Collection. 3. Evaluation tools: In evaluating usability, the following research tools were used: an- droid smart phones with a screen size of 5.5 inch; a development mobile phone ap- plication (dial screens); and a usability evaluation of proposed user interface designs on mobile phone and social media applications by a smaller cohort of elderly users. 4.2 Population and samples The sample size for the development of the fuzzy model was 200 elderly people. This was determined by factor analysis that suggested that the sample size should be at least five times the number of variables [42]. The target population were people aged 55 and over who resided in Thailand in the Mueang District of Nakhon Ratchasima Province and in the Bangkok area. Snowball sampling was used, and the participants had to voluntarily provide their personal information. A survey method was employed, combined with interviews to put participants at their ease. The usability testing sample totaled 25 elderly users, 11 males and 14 females who were selected by purposive sampling that specified that the samples needed to be people between 60-69 years of age with impaired vision conditions who had experience of using a mobile phone for one year or more. In this regard, too, the participants selected for this study were to voluntarily provide their personal information. These elderly par- ticipants had the following specific characteristics: they were aged between 60-69 years old; they had the experience of using a mobile phone for a period of one year or more; and they had eye conditions that impaired their vision. 4.3 Development of the fuzzy model The development of the fuzzy model was divided into three steps: preparation of input data; development of the fuzzy model; and evaluation of the fuzzy model for preciseness and accuracy prior to it being used for designing the mobile phone interface for the elderly users. In defining the groups of variables required for the model, input data was obtained from a feature selection of demographic characteristics and human factors as outlined below. To determine membership patterns, 14 variables were se- lected by dividing these factors into two parts: clear data of nine variables; and fuzzy data of 5 variables. Output data, on the other hand, was obtained from the program data and components of mobile phone applications and social media applications to set the membership patterns of 25 variables that could be divided by their usage into five dis- play screens, including: 8 variables of dial screen; 5 variables of call logs screen; 4 variables of chat room screen; 4 variables of timeline menu screen, and 4 variables of post screen. The data used to generate the input set for the model was obtained from a collection of demographic and human factors of the 200 elderly people recruited as described in section 4.1 and 4.2. Overall, this set consisted of 15 data inputs or variables consisting of age (X1), biological sex (X2), education (X3), eye problems (vision condition (X4), spectacles wearing (X5), eye diseases/disorders (X6), blue vision (Cyanopsia) (X7), iJIM ‒ Vol. 17, No. 02, 2023 123 Paper—The Development of a Fuzzy Model and Usability Test of a Recommended Interface Design for… color discrimination (X8), and eye measurement (X9)), screen size (X10), experience of using mobile phone (X11), technological experience (X12), memorization tech- niques/methods (X13), and memory (memory efficiency (X14), and memory effective- ness (X15)). The set excluded hearing conditions since the data collection showed that there were only two participants diagnosed with hearing disorders. The factor of hear- ing was eliminated from the data set for greater conciseness of the research and so that the specified codes met the research objectives in relation to visual impairment. The output variables, as presented in Table 1, were set with code using the Correla- tion-based Feature Selection (CFS) method with the Waikato Environment for Knowledge Analysis (WEKA) version 3.6 to select criteria from the input variables of demography and human factors that were an appropriate fit for the output variables, that is, for the application and components of mobile phone and social media applica- tions. Table 1. Output variables of components and elements of interface design on mobile phone and social media application The development of the fuzzy model consisted of the following four steps: 1. A fuzzy membership function was used to determine the numbers of fuzzy members of the input variables and the output variables obtained from the data set preparation. A cluster analysis using the K-fold cross validation technique was used by identify- ing K = 10. Also, Hartigan’s rule was used to determine the appropriate number of groups of fuzzy members [43]. This ‘rule of thumb’ suggests that ‘when clusters are well separated, then for K