Digital health (DH) technologies such as mobile apps, wearable devices, and electronic health records have potential to improve healthcare outcomes by supporting efficient workflows, reducing stress for healthcare providers, and helping people manage chronic conditions more effectively. These benefits matter particularly for older adults, who often, in highly digitized societies like Sweden, depend on DH services frequently and under demanding circumstances. However, DH solutions can lose their value in practice when usability and accessibility are poor. Common problems, such as poor text readability, cluttered features, limited language support, and interactions that require speed or precision, can lead to errors and frustration, especially for older adults or those with reduced motor control. When such issues occur, DH can become less inclusive, which can limit adoption and reduce the real-world impact of otherwise well-intended services.
This thesis investigates how usability and accessibility evaluation can be made more effective and efficient, with a focus on the practical challenges of involving key target user groups, especially older adults and people with impairments. The thesis builds on interviews with DH software professionals, which highlighted limited resources and inefficient evaluation as recurring obstacles. To study these challenges further, we conducted a systematic literature review of DH usability evaluation methods and tool support, an online survey with DH usability experts on current practices and efficiency threats, and empirical, data-driven studies that begin with field studies of older adults with shaky hands performing smartphone interaction tasks and progress toward design-time imitation of touch behavior during the design of UI prototypes.
Across the studies, usability evaluation emerged as one of the key challenges in effective DH development. The survey results indicated that participant recruitment is often perceived difficult, the time and budget are often insufficient, together with gaps in knowledge about method selection and tool familiarity. The literature also showed that usability evaluation automation support is still limited and is mainly used for user performance tracking and to check compliance with accessibility guidelines. To address recruitment limitations, the thesis demonstrates how collected data on GUI interactions can be augmented with synthetic data generation approaches and then used to train imitation models of older adults’ tapping and gesture behaviors. This resulted in the User Interface Interactions Imitation (UI3) framework, which integrates “virtual older adult users” into design prototyping environment to simulate interaction patterns and surface potential usability and accessibility issues.
The thesis contributes a consolidated evidence on where and why usability evaluation, from a research and expert perspective, often becomes inefficient in practice; an overview of the current limitations in test automation and why advanced tool support is still uncommon; and a data-driven, AI-based approach that can complement traditional usability testing through design-time imitation of older adults’ smartphone interactions in UI prototypes, enabling earlier identification of usability barriers.