Chennai: Researchers at the Indian Institute of Technology, Madras, have developed a virtual reality (VR) and artificial intelligence (AI)-based system that can identify with reasonable accuracy whether an 11 or 12-year-old child needs academic support, potentially weeks or months before the need would normally become evident through school report cards.Unlike conventional school tests that assess whether an answer is right or wrong, the new system analyses how children arrive at their answers by capturing behavioural cues such as response time, number of attempts and types of mistakes they make while completing short academic tasks in a virtual environment. The findings, published in the journal Child Neuropsychology, suggest such behavioural data could help schools identify learning gaps at an earlier stage.The study involved 120 school students aged 11 to 12 years who wore a VR headset and completed a series of tasks, including simple mathematics problems, reading a clock, counting dots and building words from scrambled letters. The assessment took around 15 minutes, during which the system recorded details not typically captured in classroom tests, including response time, attempts and error patterns.The data was analysed using machine learning models to identify patterns linked to academic performance. Using teacher assessments as the benchmark, the best-performing Random Forest model predicted whether a child would be rated below average, average or above average with 95% accuracy.“What we have built is a measurement tool. It is frugal by design, which matters if you want something usable in schools across the Global South, not just in a lab. The bigger goal is to catch learning gaps early,” said Prof M Manivannan, who also heads the experiential technology innovation center (XTIC) at IIT Madras.The research was carried out by research scholar Mridula T V and Prof Manivannan from the department of applied mechanics and biomedical engineering.The VR tasks were based on Swiss psychologist Jean Piaget’s theory of cognitive development, which considers ages 11 and 12 a key stage when children transition from concrete to abstract reasoning. Researchers also found that response time was a strong indicator of how comfortable a child was with the material, in addition to whether the answer was correct.Prof Manivannan said the team eventually wants the system to identify subtle indicators such as facial micro-expressions and involuntary movements that could point to learning difficulties, including dyslexia, dyscalculia and ADHD, enabling schools to provide support at an earlier stage.
