MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641105080 A) filed by Hitesh Kumar S; Jayasuriya D; and Shalini M on September 01, 2026, for Performance Optimization Of Game-Derived Behavioral Data For Early Dyslexia Screening Using Gradient.
Inventor includes Shalini M.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: This invention relates to a machine learning—based screening system designed to automatically evaluate behavioral data collected from children playing a language- independent visual and auditory computer game, for the purpose of early identification of developmental dyslexia risk. The system processes gameplay-derived measurements such as click timing, response accuracy, and auditory recall performance, and classifies each child as being at low risk or high risk of dyslexia, along with an associated risk probability score. The implementation makes use of a two-level stacked ensemble architecture, termed the Tri-Boost Stacked Ensemble (TBSE), in which multiple gradient boosting algorithms operate as independent base learners whose outputs are combined by a meta-learning classifier to produce a final, more reliable prediction. Dataset preparation involved cleaning of raw gameplay logs and reduction of the feature space using a tree-based feature-ranking method, and class imbalance between diagnosed and non-diagnosed cases was addressed through synthetic oversampling. The platform supports evaluation through repeated cross- validation to ensure robustness on limited sample sizes, and is designed to operate on standard consumer-grade computing hardware without requiring specialized clinical equipment. Generated predictions can be visualized through an interactive dashboard, allowing parents, educators, or clinicians to view a child's risk classification along with an explanation of which behavioral factors contributed most strongly to that result. This invention offers a practical and accessible solution for early, language-independent, and explainable dyslexia risk screening.
Disclaimer: Curated by HT Syndication.
