Integrating User Analytics into Mobile Applications for Speech and Language Treatment in Autism

Background:  Today it is common for Internet-based businesses to collect data regarding customer behavior and exploiting such behavior to expand their business. However, more introspectively such companies also seek to use these data to understand their effectiveness in serving their customers. In fact, the translation of research efforts in autism technology almost demands these same types of user analytics as rationale for those who would invest in and seek to promote the commercial enterprises that are essential for the translation of such research into practice. Specifically, translation necessitates the confluence of at least 4 key elements i) a societal recognition of a problem, ii) a common understanding of the economics involved in addressing the problem, iii) proof that delivery of a solution to the problem is accepted in the marketplace, and iv) protectable intellectual property that allows an enterprise to recover the costs of translating the research into practice in the market.

Objectives:  The objectives are to address the double bottom line problem: how to translate research into practice in a manner that is economically feasible and that simultaneously provides societal benefit. The former activity is addressed by assessing the societal value of delivering more effective mobile technology solutions to the autism population, while the latter focuses on measuring the degree to which this activity has the potential for impact.

Methods:  Autism tools created for today’s popular tablet architectures (e.g. iOS, Android) can be fitted with common usage analytics collection tools (e.g., Google Analytics) as a first order usage information collection mechanism. We will illustrate this scenario using an application for augmentative and alternative communication (AAC) training in minimally-verbal autism: The SPEAKall!® application has been instrumented with Google analytics to measure communication activities performed by the population using SPEAKall!. This instrumentation allows the collection of usage patterns from a large population of AAC users, which is distinct from the assessment of individual effectiveness in and immediately after therapy delivery sessions. While collecting such usage patterns does not suffice as evidence for the effectiveness of a therapeutic approach, it does demonstrate the market acceptance and employment of an approach in practice, and is a completely objective measure of mobile technology usage intensity.

Results:  Preliminary results indicate the SPEAKall! application was downloaded by 9,223 users over a 6-months period; 12,810 therapy sessions took place with an average length of 11min. A bi-modal weekly therapy pattern emerged. Most users composed 2-word utterances, more than half proceeded to 5-word sentences. Google Analytics also revealed most frequently used interface features and motor patterns for accessing app content. Further data analyses will examine vocabulary growth by studying the degree to which users create new symbol vocabulary. Finally, we will also present a study of geographical user spread.

Conclusions: App analytics technology can be a meaningful way to assess and justify implementation of mobile technologies in autism treatment. It generates an additional strand of evidence for technology solutions beyond traditional efficacy data from behavioral or neurophysiological investigations.