AI Accessibility Assistant

Context

During the final phase of my master's degree, I conducted research into how healthcare and welfare organisations can improve their digital accessibility. The end result is a prototype in the form of an AI accessibility assistant.

Due to the increase of digital tools in healthcare, it is important that both healthcare providers and healthcare users have sufficient digital skills to deal with this. However, according to research, a quarter of all healthcare users in the Netherlands lack the knowledge and skills to use digital care. This development requires, among other things, that digital services of organisations in healthcare, health and welfare are designed to be understandable and accessible.

The aim of this project was to develop a prototype for a tool that supports organisations in healthcare, health and welfare in assessing the digital accessibility of their applications.

How can a tool support healthcare and welfare organisations in improving the digital accessibility of their platforms?

Research

The research was conducted according to the Double Diamond model and began with research into understanding the context of the problem and user needs through interviews, literature research, stakeholder mapping and competitive testing.

Key insights 

Various target groups

The interviews conducted during the exploratory phase revealed that the target group is highly diverse in terms of background, knowledge and digital skills. For example, a general practitioner responsible for the website indicated that he did not know whether it was accessible or how to check this.

Accessibility knowledge

Although there is motivation to improve accessibility, the necessary knowledge about digital accessibility is often lacking. In addition, time pressure in the healthcare sector is a major obstacle.

Concepting

Next, ideas were generated and tested for potential solutions to the problem. Methods such as sketching and brainstorming were used for this purpose. The generated methods were evaluated using methods such as affinity mapping and SCAMPER. This was done on the basis of the values of accessibility, inclusiveness, explainability and efficiency.

Solution

The ultimate solution is an AI Accessibility Assistant that examines the platforms of healthcare and welfare organisations. The assistant collects data and the model scans it using a combination of natural language processing, computer vision and code checkers. The model assesses the data based on accessibility guidelines and design heuristics. The results are customised to the user and presented using storytelling to teach the user about accessibility, thereby encouraging the user to address the problem. 

Several iterations of prototypes have been designed, tested several times and evaluated. 

Reflection

The research has shown that healthcare and welfare organisations increase the accessibility of digital tools in the healthcare and welfare sector through personalisation and storytelling. Storytelling conveys knowledge and strengthens both understanding and motivation to tackle accessibility issues, while an experience-based explanation choice promotes inclusivity.