A considerable number of students are affected by different types of learning disabilities. These learning disabilities affect student’s educational progression in life, students get less motivated and pass through frustration and low self-esteem. Due to a lack of awareness amongst the community majority of these students remain undetected, because of this, they struggle in continuing education and are unable to get special required education. Early detection of learning disabilities plays a major role in the upliftment of these children and helps them to provide a remedial solution. This paper focuses on the discussion of different intelligent approaches that work for the prediction of learning disabilities and proposes a handwritten text-based approach for the early prediction of learning disability mainly dyslexia. Here, various approaches like prediction of dyslexia using handwritten text, gaming approach, eye tracker, and prediction using brain and EEG approaches are discussed. Amongst these approaches, it is found that the prediction using handwritten text is the most promising, scalable, and accurate. Children having learning disabilities like dyslexia possess difficulty in processing similar-looking characters like b-d, 2-z, etc. Also, these students have dissimilar handwriting patterns compared to normal children. Using handwritten text recognition and deep neural networks, predictions can be made about whether children are normal or dyslexic. The preliminary experimentation is performed with the use of the IAM data set of writing features using 3 different assignments given to children and patterns are extracted from these texts.