MindKey
An assistive text prediction keyboard system for gaze-enabled AAC devices
MindKey is a personalised text prediction keyboard system designed for gaze-enabled Alternative and Augmentative Communication (AAC) device, adopting machine learning language models. I was honoured to be able to work with two MND patients, Alan and Mike & Mary, during the design process.
Project Overview
The project scope was initially based off my own interest — to facilitate communication in MND patients who have lost their voice.
User Research
To better understand the user needs, pain points, goals, and everyday challenges of MND patients, I carried out secondary research via MND forums, anecdotes, and existing assistive technologies for the MND community. I also conducted semi-structured interviews with Alan and Mike & Mary via Zoom to gain first-hand insights.
User needs, pain points, goals, and motivations of research information were developed through secondary inductive thematic analysis, which guided the development of a user journey map of depicting a patient’s progression regarding loss of speech ability and the use of existing assistive technology for communication.
Design Ideation
I based my design off an existing assistive keyboard system for AAC devices, the “Grid 3 Fast Talker System”, as well as published literature on existing language models.
User Requirements
Low-fidelity Sketches
Initial High-fidelity Design
Prototype Evaluation
Example keyboard interface with good eye-tracking quality.
Final Design
Implications & Contributions
My Learnings
