Abstract
Users with cognitive impairments use an assistive technology email system, CogLink, for socialisation and help in their activities of daily living. As users interact with the AT email client, the logged stream of events is monitored and analysed to aid decision-making. When caregivers receive monitor notifications, they know that the user has had a significant change in her emailing behaviour. Consequently, caregivers select adaptations to the email client that can challenge the user to gain still more emailing skills. The monitor in this emailing system analyzes user data to recognise when significant changes in user behaviour warrants caregiver attention. Moreover, the monitor can distinguish newly learned skills from new, but transient behaviours. This article summarises the CogLink assistive technology monitoring techniques, introduces the learning likelihood algorithm, which distinguishes transient from learned behaviour, and provides lessons learnt from a decade of monitoring CogLink users.