Learning disorders (LDs) are neurodevelopmental disabilities with a worldwide prevalence of 5-15%. Lack of awareness paired with heterogeneity in testing methods results in non-identification of LDs in children. Assessment tests presently used to diagnose LDs require the physical presence of a medical professional, are time-consuming and expensive, and adopt a non-child-centric approach. DysDiag proposes novel, accessible, and easy-to-administer risk assessment and screening tests (based on DSM-5 criteria), for LDs in children (5-8 years). DysDiag’s test for Dyslexia consists of a gamified, visual-based quiz that accesses the child’s phonemic, auditory, and visual-based skills followed by a pronunciation test and parental questionnaire. The test for Dysgraphia includes 2 Machine Learning Image Classification Models that classify the child’s handwritten sample as dysgraphic or normal and further evaluate the sample for 6 diagnostic symptoms. The models recorded F1 scores of 0.785 and 0.964 respectively. The test of Dyscalculia includes a facial emotion recognition model alongside a response-time-based math quiz and a parental questionnaire. DysDiag was tested on 40 children consisting of a case group of pre-diagnosed children (n=20, mean age=6yrs) and a control group (n=20, mean age=7yrs). The children were tested by a registered medical professional followed by DysDiag’s screening tests. DysDiag recorded a sensitivity and specificity of 90%, a Positive Predictive Value of 94.73%, and a Negative Predictive Value of 90.47%. DysDiag was also reviewed and rated by 15 psychologists and pediatricians. DysDiag proved to be a clinically viable tool that can aid in the early identification and mass screenings for LDs at elementary schools.