One million adults are diagnosed each year in the United States with cognitive impairments (CI). Many of these individuals are unable to travel unassisted outside their residences to participate in community or social activities, which results among other things in social isolation for this growing segment of our society. While assistive technology (AT) holds great promise for helping people with CI achieve independence, studies show these users frequently abandon AT systems due to (an eventual) misalignment between personal user goals and abilities, and the functionality delivered by the system. Whereas unimpaired users may learn and adapt their behavior to exploit system functionality, loss of cognitive abilities makes this infeasible for CI users. Previous work by the PI in developing AT for CI users has shown that the problem of device abandonment can be mitigated by personalizing AT design to the particular goals and capabilities of the user, and then re-designing as those capabilities change over time. One result of this prior research is a framework for assessing and characterizing the personal and contextual requirements (PC-RE) of the user. The PI has applied this model to build personalized systems that align with user goals, support monitoring for long-term requirements discrepancies, and guide re-design. Longitudinal studies have shown that this "assess-personalize-build-monitor-adapt" approach is effective for representative CI populations, but unfortunately it is largely ad hoc and clearly cannot scale to provide the large numbers of personalized systems needed by the growing CI population. In this project the PI will address these deficiencies by developing and formalizing a novel design approach that supports rapid, cost-effective mass-personalization of AT systems. The approach provides a new software development mode the PI calls a Software Pharmacy. Operationally, a Software Pharmacy takes a formal PC-RE specification (a prescription) embodying the goals, requirements, and skills of an individual user, and from it generates a self-monitoring, self-adaptive set of system software meeting those needs. An underlying development model based on techniques from social science, PC-RE, software product-lines, and dynamic requirements monitoring will provide the capabilities supporting rapid production, re-configuration, and re-design in a feedback/control development cycle. The PI will formalize the development model, operationalize it for a set of travel-assistant devices, and empirically test effectiveness of the design paradigm in a set of longitudinal studies with CI users. Resulting models, methods, exemplars, and tools will be transferred to the Design Science and AT development communities.
Broader Impacts: Researchers currently lack both methods and tools supporting effective, low cost software personalization. This work will provide design methods, models, examples, and tools supporting end-to-end development of personalized AT system software. These artifacts as well as the theory, examples, and data from experimental application will support Design Scientists in addressing a wide range of emerging problems in creating software systems for individuals who currently lack access (user-appropriate technology) or demand systems fitted to their individual capabilities and needs. Results will also be useful to commercial developers for developing new systems that better support social integration for people with CI.
