Deep learning (DL) based improved probabilistic dense model (IPD) for autism spectrum disorders (ASD) classification analysis

Abstract

Among other symptoms, people with autism spectrum disorder (ASD) suffer from communication, social interaction, and repetitive habits. Even while complete eradication is not common, early interventions may make the illness more manageable. Here, we propose a practical framework for contrasting various Machine Learning (ML) methods for diagnosing ASD at an early age. The proposed framework uses a number of Feature Scaling (FS) methodologies, such as min-max scalar, principal component analysis, and intuitive visual representation analysis, to incorporate an efficient prediction method into the overall architecture. Our recommended design, which utilizes the IPD model, suggests that the categorization of adult autism spectrum condition, as an entity, is reflected in the overall verification of the features, and varied functional qualities. To do the mathematical analysis in our IPD model, we use chi-squared and probability density functional methods. The structure's purpose is to recognize and classify patients based on their palpable traits. We can observe that the 95.9% accuracy is far greater than the Machine Learning approach in the end.