Clinical Trials to support autistic people

Autistic people often face significant barriers in communication and in understanding others’ emotions, yet current support tools remain limited. This programme seeks to evaluate two complementary forms of assistive technology designed to improve everyday communication and social interaction in real-world settings.

Project A: Communication Support for Non‑Speaking Autistic Children

The first component focuses on autistic children who are non speaking or minimally speaking. Approximately 30% of autistic people fall into this group, but most still rely on older, low tech communication systems such as PECS (the Picture Exchange Communication System) or traditional speech and language therapy. We will conduct the first randomised controlled trial of Proloquo2Go, an Augmentative and Alternative Communication (AAC) app for iOS devices. The app provides symbol based and text to speech communication, with over 27,000 symbols, customisable vocabulary grids and bilingual voice options. The app provides symbol based and text to speech communication, with over 27,000 symbols, customisable vocabulary grids and bilingual voice options. It provides over 100 natural-sounding text-to-speech voices in multiple languages and accents, including children’s voices. Its Crescendo™ Vocabulary system allows children to begin with a small set of core words and expand as skills develop. We will recruit 200 autistic non speaking people: 100 will be randomised to use Proloquo2Go and 100 to routine speech therapy. Groups will be matched for age, IQ, gender, parental education and co-occurring conditions. The intervention will run for one year, with blinded assessments at baseline, 3, 6, 9 and 12, 18 and 24 months.

If the results are positive, we will prepare policy guidance to support the implementation of these devices across clinics and schools. Analyses will also incorporate the autistic subgroups identified in Programme 1, enabling us to establish whether specific profiles benefit more from this type of support.

Project B: Smart‑Glasses Support for Mental‑State Recognition
In the second component of this programme, we will evaluate whether wearable technology can help autistic people interpret others’ emotions and mental states more accurately. Many autistic individuals find “theory of mind” tasks challenging, which can affect social interactions.

Working with leading technology experts, we will integrate machine learning algorithms into smart glasses capable of identifying facial expressions and mental states in real time. Using our lab’s validated set of 412 mental states, Mindreading, spanning a broad range of emotional intensities and valences, we will test how accurately the glasses classify mental states under repeated and varied conditions compared to human consensus labels. Each stimulus will be presented to the smart glasses system under controlled laboratory conditions.

The embedded camera and on-device emotion recognition algorithms will process each video or image frame to generate mental-state labels.

This phase aims to achieve a device with high reliability, validity and consistency before moving to clinical trials. Once the technology reaches the required accuracy threshold, we will conduct a full RCT with 200 autistic adults (equal numbers of men and women), randomised to wear the smart glasses or to a control group. Subsequent trials will extend to younger age groups.