Assisting the Design of Consumer Internet of Things Applications Using User Experience Patterns Incorporating Machine Learning

User experience (UX) design has become a pervasive function in the modern digital product lifecycle, spanning initial concept through to post-launch optimisation. Within the consumer Internet of Things (IoT), continuous streams of behavioural and contextual data present both opportunities and challenges for UX professionals willing to craft more intelligent and adaptive experiences for endusers. Machine learning (ML) is a typical mechanism leveraged to take advantage of such a data availability, yet many designers lack the literacy required to specify MLdriven interactions. This doctoral research explores an approach, grounded in Alexandrian Pattern Theory, that enables UX designers – regardless of their level of ML literacy – to conceptualise feasible adaptive interactions for the consumer IoT. The work begins by formalising the Personal Spark methodology pattern, which combines engagement with field experts and the “What Would a Human Do?” thought experiment to uncover and refine recurrent design knowledge. Employing Personal Spark in one-to-one workshops with UX professionals (N = 23) uncovered five interconnected UX patterns – Preference Learner, Geographic Trigger, Smart Shortcuts, Stop Learning Preferences, and the proto-pattern Intelligent Access Sharing – collectively forming a novel, extensible UX pattern language for consumer IoT that integrates ML. A mixed-methods, proof-of-concept evaluation enlisted UX designers (N = 5) who, after approximately ten to fifteen-minute briefing on two patterns (Preference Learner and Geographic Trigger), each produced a comprehensive consumer-IoT user-flow concept within an average of twenty minutes, all of which were deemed technically feasible in an independent IoT-expert assessment. The principal contribution to UX scholarship and practice lies in operationalising Alexandrian Pattern Theory for data-driven contexts, therefore providing a shared language that aids the collaboration between UX design and allied functions whilst reducing the expertise needed for the integration of ML into UX solutions.