FAU KIMALAYA project for analyzing images receives 179,000 euros in funding

Symbolic picture for the article. The link opens the image in a large view.
Image: AdobeStock/contrawerkstatt

Objective school entry health examinations thanks to artificial intelligence

Before starting the first grade of school, children in Germany undergo a school entry health examination to check their physical, mental and social-emotional development. The aim is to prevent children from starting school too early and to detect any needs a child may have early on. Some parts of this examination can, however, depend on the subjective viewpoint of the person conducting it. A new analysis method developed by Prof. Dr. Andreas Kist, Professorship for Artificial Intelligence in Communication Disorders at FAU, aims to increase objectivity and consistency during the examination thanks to artificial intelligence. The project has received funding from the Bavarian State Ministry of Health and Care (StMGP).

The purpose of the school entry health examination is to test, among other things, the vision, hearing, height and weight and motor skills of children who are about to start school. The scores given in numbers are easy to compare and are unambiguous. Assessments of speech development and vocabulary do not allow much room for interpretation either as the child being tested either knows a certain number of words or does not.

The analysis of social and emotional skills, which show how children behave in their environment and how they deal with their emotions, is a different matter, however. The expectations and experience of the person conducting the test can sometimes greatly affect the results in this case: While one person may regard a child as shy, another may think they are just quiet.

Using standardized methods to examine children’s drawings

Prof. Dr. Andreas Kist’s research project, funded by the Bavarian State Ministry of Health and Care (StMGP), could increase the objectivity of these examinations. KIMALAYA, short for “KI-basierte Analyse von gemalten Bildern im Rahmen der Schuleingangsuntersuchung” (AI-based analysis of drawings as part of the school entry examination), is investigating the use of artificial intelligence during the school entry examination. The main focus of the project is the standardized detection and systematic analysis of children’s drawings.

Drawings or paintings made by children reflect their creativity, cognitive abilities and personality. They can provide valuable insights, especially when a child’s verbal skills are not extremely well developed or when the standardized test does not match the situation.

Sometimes, the drawings of one child are not very similar to each other, even if the requirements remain the same. AI should be able to take these variations into consideration.

AI analyzes analog examples

Prof. Kist uses analog drawings made on paper for his analysis. In contrast to apps or other digital drawing programs, drawing on paper using crayons or pencils has several advantages besides its accessibility: Children can demonstrate their motor skills and their hand eye coordination and there are no technical barriers or delays. “In this case, digitalization does not mean that children have to complete every task on a computer or tablet, but that the results are recorded digitally,” says Prof. Kist.

Future milestone: KIMALAYA on smartphones

To enable this digitalization, one core aspect of the KIMALAYA project is to allow the AI-based analysis to be carried out on a smartphone by the person conducting the school entry examination. Prof. Kist is aware that this will be challenging: “It’s one thing to develop robust and very precise AI solutions that operate on powerful computers,” says the AI expert. “However, optimizing these methods for a smartphone without making significant losses in the quality of the AI is a completely different matter.”

KIMALAYA starts at the beginning of 2024

As part of the KIMALAYA project, Prof. Kist has been training an analysis program supported by AI to examine drawings for indicators such as geometry, proportions or how human anatomy is represented. The aim is to integrate the system into a smartphone app for use during the school entry health examination.

A traffic light system in the KIMALAYA app will display the system’s evaluation of the child’s developmental level in this area. The project will run for 19 months, during which Prof. Kist will be supported by 179,000 euros of funding from the Bavarian State Ministry of Health and Care (StMGP). The project is based in the Department of Artificial Intelligence for Biomedical Engineering (AIBE) at FAU, which is supported to a significant extent by the High-Tech Agenda Bavaria.

Further information:

Prof. Dr. Andreas Kist

Professorship for Artificial Intelligence in Communication Disorders