THE PROMISE OF USER-CENTERED DESIGN IN THE CREATION OF ASSISTIVE TECHNOLOGY
Dr. Grace Teo, founder of the Open Style Lab and lecturer of Principles and Practice of Assistive Technology (PPAT) at the Massachusetts Institute of Technology.
User-centered design is an emerging approach employed in the design of assistive technologies. Its use is driven by the wide diversity of disabilities, which often necessitates customization to the specific needs of each user. By beginning with the individual needs of a user, we can ensure that the resulting technologies are suitable for the user’s needs, desires, lifestyle and environment. This is particularly important in assistive technology, where rates of technology abandonment have been reported to be as high as 30%1. Indeed, one of the top predictors of assistive technology abandonment is the lack of consideration of user opinion during the technology selection process1. By encouraging user participation in the technology selection process, rates of abandonment can be greatly reduced2.
The value of understanding the unique needs of each user is strongly recognized in the field of occupational therapy. Occupational therapists work closely people with disabilities to help them perform activities that are part of daily living, e.g. eating, personal grooming and using a computer. Often, they must select and/or modify assistive technologies for their patients, to enable them to achieve specific tasks. Several models of assistive technology selection exist that share a commonality in trying to understanding the impact of (i) the user’s abilities and values (the ‘Human’), (ii) the requirements of the desired task to be performed (the ‘Activity’) and (iii) the social, cultural, environmental and institutional factors (the ‘Context’) on the selection of assistive technology. Ideally, the assistive technology should suit the Human, Activity and Context to increase the likelihood of user success. If no available technology exists to fit the need, then customized assistive technology might have to be designed and created.
Creating customized assistive technology from scratch often calls for advanced engineering skills, with mechanical engineering, electrical engineering and software programming skills being key among them. Unfortunately, apart from human-computer interaction in software engineering, user-centered design is not emphasized in most engineering fields. Perhaps this is due to the difficulty of characterizing assistive technology systems in engineering terms. Often, engineered systems are rigidly characterized by equations that determine how an input into a system translates into output. People and natural or socio-cultural environments are much less amenable to such characterization, however, and might thus be avoided by engineers. Nonetheless, the functional, success metric-oriented approach of engineering can still provide overall guidance in setting specific, measurable goals for the development of assistive technology.
On the other hand, designers (especially interaction and industrial designers) have a broad arsenal of tools to tackle the more amorphous challenge of human-centered design. In particular, iterative design methodology helps to accommodate the multiple, unexpected factors in real-life user scenarios that might be difficult to comprehensively predict and plan for. In iterative design, designs are prototyped and tested across multiple cycles, with the quality or ‘fidelity’ of the prototype improving in each cycle. For example, the design of a service dog cart to help a person with chronic fatigue with grocery shopping might have early prototypes that are made from cardboard and tape, with the aim of gauging whether the width of the cart is suitable for maneuvering through the client’s neighborhood and grocery store aisles. Later prototypes might use high quality materials and more advanced assemblies, e.g. aluminum and screws, to test the durability and strength of the cart. In each cycle, the user is invited to test the prototype, with the hope that inadvertent design flaws can be caught early on, before too many design details are committed to.
Combining the complementary strengths of occupational therapists, designers and engineers has much potential for designing successful assistive technology for people with disabilities. An example of multidisciplinary, user-centered design in the creation of assistive technology is the Open Style Lab platform. Occupational therapists, designers and engineers (the ‘fellows’) are teamed to create a functional yet stylish wearable solution with and for a client with disability facing a clothing challenge (the ‘client’ or ‘user’). Importantly, the user with a disability initiates the project challenge. Hence, user-centered design is incorporated from the beginning, with fellows exploring the client’s abilities, activity needs and contextual factors that will affect the success of their solution.
On the practical level, fellows are given time to spend with their clients to better understand the scope and subtleties of the challenge, primarily through contextual inquiries. While traditional interviews can allow for efficient exchange of ideas and information between fellows and clients, observation is crucial for a thorough understanding of the client’s lifestyle and experience, and often yields unexpected but important insights. Contextual inquiries are a combination of traditional interviews and observation. A ‘master-apprentice’ model is used, where fellows act as ‘apprentices’ to learn and observe how their clients currently tackle their clothing challenges. Importantly, the client’s responsibility is to demonstrate how they tackle their clothing challenge in context, explaining the rationale behind each step, as if they were training the participants to replicate their actions. Borrowed from the field of ethnography, contextual interviewing is a useful technique to increase empathy with the client and understand a client’s needs without being interrogative. Respecting the client’s expertise in his own needs also encourages the client’s agency within the design process, setting the stage for increased participatory design in subsequent stages, e.g. user testing.
At this early stage of defining the client’s challenge, fellows must be encouraged to check their assumptions repeatedly, as a false understanding of the challenge will necessarily lead to an inaccurate solution. For example, we have previously encountered a client with cerebral palsy who wore a head pointer at work to use a keyboard and mentioned that this solution was tiring. The designers and engineers on his team initially presumed that his fatigue was localized to his neck, and recommended his transitioning to an eye-tracking device instead, to prevent use of his neck. However, during testing, the client surprisingly scored neck fatigue as equal for both the head pointer and eye-tracking device. Upon further questioning, it turned out that the client’s head pointer fatigue was actually localized to his eyes, which had to repeatedly look up and down between the keyboard and computer monitor! Thus, clarity on the root causes behind symptoms that clients report is extremely important, and hinges on the effectiveness of communication between the client and fellows.
Compared to the initial stage of understanding and defining the client’s challenge, incorporating user-centered design principles in subsequent design and fabrication stages is far more ambiguous. Specifically, how does user-centered design influence the allocation of responsibilities and authority between the user and fellows to make design decisions? For example, in making a temperature-regulating jacket, a decision to create an automated feedback system or a manually adjustable system might have to be made. The client might strongly desire an automated feedback system, the engineer on the team might have technical concerns about the feasibility of an automated system, while the occupational therapist may have fears of overheating and burn injuries with both systems. Personalities with different risk tolerances, diverse occupational expertise and the felt obligations to follow the client’s wishes can conflict. Further, clients with both physical and intellectual disabilities often have long-term caretakers who express needs on their behalf. Hence, it is often challenging to assign weights to the differing opinions of all involved when controversial design decisions must be made.
One way of resolving this issue is to give fellows the responsibility of educating the user in the scope of possible solutions, and the benefits and risks associated with each option. The success of this method is contingent on the fellows’ abilities to comprehensively and clearly describe the options, and the clients’ abilities to make a rational and utility-maximizing decision given the information provided. Fellows need to possess advanced communicative skills, such as using non-technical language to describe complex subject matter, and the ability to draw diagrams to aid comprehension3. On the other hand, various traits, such as self-awareness and ability to make accurate judgments4, will influence clients’ abilities to make decisions that most benefit them. Further, research suggests that in clinical care settings, patients are more risk averse when they must take responsibility for their own care decisions5. In reality, many clients choose to allow experts to make decisions on their behalf. Indeed, in one study 69% of patients with chronic disease preferred to leave medical decisions to their physicians6. Given these considerations, the most feasible and practical way forward may be a middle ground where fellows communicate only the most significant decision splits, and give clients the authority to allocate decision-making power between the fellows and themselves.
Perhaps the most pertinent underlying question we should ask ourselves is this: Does user-centered design increase acceptance of assistive technology because considering user input creates an optimized design, or because considering user input gives the user more ownership in the final product? The answer is likely different depending on the client. The more inclined the answer is toward the latter scenario (i.e. ownership is key), the less important it might be to include the user’s input in all facets of the decision making process, but the more important it is to give them the choice regarding which facets they would like to be involved in and how they wish to be involved.
In some instances, clients may not have the capacity to choose. The former is particularly true for clients with intellectual disabilities. Here, long-term caregivers – a parent, personal care assistant, or skilled nurse – might have to make a decision in the client’s stead. Again, communication and rational decision making skills will be key for fellows and the caregiver. To note, much has been written on the topic of the ethics of surrogate decision making, particularly in the context of advance directives and end-of-life care. While this is outside the scope of this article, the alternative decision making models that have been discussed in such contexts should be considered. Specifically, the popular model of ‘substituted judgment’, where caregivers attempt to make a decision that they believe the client would most likely want, has been demonstrated to be a flawed approach. Surrogate predictions are shown to be correct only about 68% of the time7, and facilitated conversations and instructional materials have had limited success in improving prediction accuracy8. Two possible alternative models have been proposed: (i) a best interest model that evaluates the benefit of decisions based on community norms and (ii) a narrative approach that tries to make decisions that seem in line with the patient’s previous life choices9. These models may be helpful in guiding design decisions where clients are not able to participate.
Beyond design decisions, fellows engage in rapid prototyping methods to test and refine their solutions iteratively with clients. To carry out user testing, success metrics must be defined, and the client’s personal notion of competence is a key determinant of the metrics. Often, a client’s notion of competence is a result of discussions with their care provider or the fellows pertaining to what a reasonable expectation for success might be. The resulting metrics are useful for quantitative measures, but become problematic in qualitative measures (e.g. Likert scales where the client rates their subjective experiences of the prototype). Occasionally, clients can be overly enthusiastic about a solution’s efficacy to preserve the team’s morale or might change their minds unpredictably regarding the design features they wish to have. Hence quantitative measures are typically preferred, if possible.
Another factor to consider during user testing with people with disabilities are issues of fatigue and flux in the health of the client. Fellows are advised to consider the client’s physical needs when planning extensive user testing sessions, take regular breaks, or even reschedule tests if needed. Moreover, a client’s physical health might evolve over the course of the design process, so in order to obtain the same functional readout over time for a prototype, the user test might also have to change to accommodate the client’s changing abilities. For example, one of our clients who was diagnosed with Amyotrophic Lateral Sclerosis (ALS), a condition that severely weakened her arm and shoulder muscles, experienced unexpected and rapid progression of her disease over the design process. Her team of fellows designed a belt that would support her arms while walking. Due to the disease’s progression, the team had to change their testing strategy, from measuring her subjective shoulder discomfort after her typical stroll in her park to measuring shoulder discomfort while walking around her home. Developing a more objective metric of shoulder support by measuring shoulder gap reduction with different arm support prototypes was also useful. Hence, user testing in assistive technology development needs to be more sensitive to the user’s health and work creatively around changing constraints.
Finally, user-centered design can result in extremely customized solutions, which must evolve for large-scale commercialization or distribution. Indeed, the path from designing for an individual’s needs to a mass-market solution is not well defined, and wrought with compromises. Here, the emergence of do-it-yourself (DIY) culture and makerspaces might offer an alternative way of scaling up these solutions. Specifically, given a set of instructions, people with disabilities and their caregivers might be able to adapt existing solutions to their own circumstances. One of the top online platforms where instructional exchange occurs is the Instructables website (www.instructables.com). The website features an assistive technology channel, with illustrated steps for making devices that range from a piano pedal that requires minimal strength for use, to a vibrating obstacle detection sensor. Complementary to these resources, makerspaces or ‘fab labs’ (short for fabrication laboratory) provide shared production tools and equipment for their members or the public to use. Typical equipment includes 3D printers, laser cutters, soldering stations and milling machines. Some examples of makerspaces are RogLab in Slovenia and Artisans’ Asylum in the USA. Indeed, many people with disabilities are very inventive and willing to create their own solutions given the lack of commercially available options. The only downside to DIY solutions is that they are often time and energy intensive, and equipment in makerspaces are typically difficult for people with disabilities to use due to the strength or dexterity needed to operate the machines.
To address the difficulty of creating assistive technology from scratch and to lower the barrier of entry to DIY fabrication, assistive technology kits might be used instead. Kits comprise a standardized set of components that can be assembled along different design trajectories to result in multiple different devices. Assembly can be designed to be easy for people with disabilities, using magnets or physical slots to assemble different components together. A good example of a kit, though not specifically designed for disabilities, is an electronics prototyping platform LittleBits (http://littlebits.cc/). Each kit contains color-coded and magnetized components that snap into place easily to make various connected appliances. Ideally, kits should further invite the user’s creativity to add components not included in the kit, to increase or modify the functionality of the end-result. Currently, there are few assistive technology kits on the market and thus represent a sizable potential commercial opportunity.
In conclusion, the use of user-centered, participatory design in assistive technology creation is yet to be optimized, but has already demonstrated potential for filling unmet needs within the disability community. Multidisciplinary teams can help combine user-centered approaches with the necessary technical skills for such initiatives. While there are no fixed guidelines for which team member should make design decisions, giving users the opportunity to choose how they wish to be involved in decision making, and how decision making should be carried out, is key. Finally, coupled with the momentum of the DIY culture, specific user-centered designs can be easily shared and adapted for a larger community, presenting an exciting possibility for allowing people of all abilities to benefit equally from today’s technological advancements.
Phillips B, Zhao H. Predictors of assistive technology abandonment. Assistive Technology. 1993;5:36-45
Johnston P, Currie LM, Drynan D. Getting it “right”: How collaborative relationships between people with disabilities and professionals can lead to the acquisition of needed assistive technology. Disability and Rehabilitation: Assistive Technology. 2014
Epstein RM, Alper BS, Quill TE. Communicating evidence for participatory decision making. Jama. 2004
Riquelme H. Do consumers know what they want? Journal of Consumer Marketing. 2001
Fraenkel L, Peters E. Patient responsibility for medical decision making and risky treatment options. Arthritis Care & Research. 2009
Arora NK, McHorney CA. Patient preferences for medical decision making: Who really wants to participate? Medical care. 2000
Shalowitz DI, Garrett-Mayer E. The accuracy of surrogate decision makers: A systematic review. Archives of internal medicine. 2006
Ditto PH, Danks JH, Smucker WD. Advance directives as acts of communication: A randomized controlled trial. Archives of internal medicine. 2001
Torke AM, Alexander GC, Lantos J. Substituted judgment: The limitations of autonomy in surrogate decision making. Journal of general internal medicine. 2008