Rodina Eladawy / Mathematics / Faculty Mentor: Pedro Maia
“This research explores the machine learning models to predict cognitive outcomes, as measured by the Wechsler Intelligence Scale for Children (WISC-III), from a diverse set of predictors including visuospatial memory test results, demographic information, and the presence of ADHD. Our study aims to understand the complex link between ADHD and cognitive performance, enhancing predictive models.
Our dataset includes records from 50 children aged 10-12, with 28 having ADHD, to analyze its impact on cognitive abilities.
In this study we using Python’s scikit-learn, we analyze features with models like Linear Regression, Ridge, Lasso, Elastic Net, SVR, Decision Tree, Random Forest, Gradient Boosting, and K Neighbors Regressor. These features include reaction time statistics, performance metrics, demographic data, and clinical diagnoses. We use adjusted R2 and mean square error for accurate WISC-III predictions, accounting for model nuances.
Our top model is a Gradient Boosting Regressor, has an adjusted R2 of 0.729 and a mean square error of 24.97.
In conclusion, we provide correlation between ADHD and cognitive abilities while also showcasing the potential of machine learning to improve predictive capabilities.
in future work, we will refine our model by identifying key features from pupil time series for cognitive analysis.”
Leave a Reply