Our experience in school, especially at an early age, often defines us and our perceptions about ourselves and our abilities. Sometimes children may have difficulty in school due to their learning abilities or development, as many young children may develop differently, have different needs and learning styles. Their learning experiences in a school play a significant part in their future decision making in terms what they dream of doing when they grow up as well as their perception of their abilities. Although a child may struggle in a particular area, it does not mean that he or she struggles for the same reason as another child who struggles in this same area. Doctors and teachers can work together with children to ensure the best quality of schooling, but sometimes certain learning abilities may go on undiagnosed. To help with this problem, researchers at Cambridge University found a novel way to use machine learning algorithms to diagnose the reason why certain children were struggling. What they found was that their algorithm found learning difficulties which did not match previous medical diagnosis, giving illuminating insights into why children struggle.
The researchers at Cambridge University worked with 550 children who were struggling in school. They did not separate children based on their diagnosis, rather they looked at the whole group holistically. Including the whole range of difficulties and diagnosis allowed the authors to look at the whole spectrum of diagnosis, as well as their overlap. The AI algorithm measured each child's cognitive skills such as listening, problem solving, vocabulary, memory and spatial visualization. The results indicated that children fell into one of the four clusters: "1) children with broad cognitive difficulties, and severe reading, spelling and math problems 2) children with age‐typical cognitive abilities and learning profiles; 3) children with working memory problems; and 4) children with phonological difficulties." The researchers found that two clusters, difficulty with working memory skills, which is linked to difficulty with math, and difficulty with working with processing sounds in words, which are attributed to the difficulty in reading comprehension, shared a link. That is to say, children who had trouble in math also had trouble with reading comprehension. This is an important insight, as past research in the field did not identify this link. This indicates, that we need to move on past rigid labels, and focus more on individualistic approach to learning difficulty. The authors indicate that this piece of research serves as a way to use more novel algorithms in machine learning to better identify and help children, and their parents, with managing and understanding learning difficulties.
