Per Bækgaard, NNF programme Reading the Reader- Credit Hanne Kokkegård, DTU Compute

AI will help people with impaired vision to read better

IT and cyber security Mathematics Health technology
With a grant of eight million DKK from the Novo Nordisk Foundation, Associate Professor Per Bækgaard will be at the forefront of developing an artificial intelligence model that decodes reading patterns in people with central vision loss and adapts the reading material to the individual person.

Edited by Hanne Kokkegård

Millions of people have impaired vision, which negatively affects their daily life in various ways. Many of them have central vision loss, in which the centre of their field of vision is blurred so much that they have difficulty reading and carrying out other activities. A new research project “Reading the Reader” will address this problem.

Based on a grant of nearly DKK 8 million from the Data Science Collaborative Research Programme of the Novo Nordisk Foundation, Associate Professor at DTU Compute Per Bækgaard will lead the development of an artificial intelligence model for determining the reading patterns of people with central vision loss with the aim of adapting reading material to each person’s needs.

The research is being carried out in collaboration with Professor Sofie Beier and her colleagues from the Royal Danish Academy - Architecture, Design, Conservation and includes computer scientists, psychologists, experts in reading patterns and ophthalmologists.

“The project involves developing a model that enables us to rapidly decode a person’s reading patterns and simultaneously set, for example, the font size, contrast or spacing between letters on a mobile phone, a tablet or other screen, so that the material is optimally adapted to the person’s reading pattern, thereby providing the optimal reading experience and improving reading speed,” explains Per Bækgaard in a press release from  the Novo Nordisk Foundation.

People with impaired vision can have difficulty in education

Reading can be a challenge for people with central vision loss. Many have to move their heads from side to side and forward and back to read the text. If the text is displayed on a tablet, a mobile phone or a computer, they also often adjust the contrast, increase or decrease the brightness or change the font size.

The many adjustments required to read a text often makes taking an educational programme more difficult for people with central vision loss and can also result in social problems. This research project is taking a completely new approach to this problem.

“Much is already being done to help people to improve their reading, but the reading material is still predominantly static. We would like to determine whether we can use modern technology to categorise the reading patterns of people with central vision loss and then use this knowledge to alter the reading material to automatically improve the reading process,” says Per Bækgaard.

Tracking eye movements

The main aim of the project is to develop a Machine Learning model that can interpret the reading pattern of people with central vision loss. The model primarily uses eye-tracking to follow a person’s eyes while they read and can thereby identify when they start skipping lines, how they move their heads to enable them to read and how their pupils expand and contract during different reading experiences.

The model should then be able to decode the reading experience of each person and thus dynamically adjust parameters, including screen contrast, font size, brightness, spacing between letters or lines and other parameters, to optimise reading.

The project also aims to use the model to improve the determination of why some people have difficulty reading.

NNF's Data Science Collaborative Research Programme

The grant for Per Bækgaard’s project is one of three grants awarded through the Foundation’s Data Science Collaborative Research Programme.

The grants are awarded annually for research projects that attempt to tackle some of the world’s major problems through interdisciplinary research and collaboration between research groups within data science, medicine, biology, plant science, biotechnology, physics and chemistry.

Projects and grant recipients

  • Reading the Reader, led by Per Bækgaard, Associate Professor, Department of Applied Mathematics and Computer Science, Technical University of Denmark, Kongens Lyngby: DKK 7,893,19
  • CAZAI: CAZyme Specificity Prediction Using AI, led by Bernard Henrissat, Professor, Department of Biotechnology and Biomedicine, Technical University of Denmark, Kongens Lyngby: DKK 14,995,489
  • MOPITAS: Multi-omics Profiling in Time and Space, led by Richard Röttger, Associate Professor, Department of Mathematics and Computer Science, University of Southern Denmark, Odense: DKK 18,039,509