Exploring the Effectiveness of Machine Learning Models in Predicting Learning Disabilities in Children: A Literature Review

The literature review analyzed various studies that used machine learning models to predict learning difficulties (LD) in children. Eighteen studies were reviewed, with the majority using neuroimaging and/or functional connectivity data. The studies used a variety of models, including support vector machines (SVM), random forests (RF), logistic regression, k-nearest neighbors (k-NN), deep learning models (including convolutional neural networks (CNN), deep belief networks (DBN), and recurrent neural networks (RNN)), gradient boosting machines (GBM), multilayer perceptron (MLP), and sparse autoencoder (SAE) models. Overall, the models achieved high levels of accuracy, with most models achieving accuracies above 80%. The highest accuracy achieved was 95%, while the lowest was 78%. These results demonstrate the potential of machine learning models in predicting LD in children and highlight the importance of incorporating neuroimaging and/or behavioral data in such models.