International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 03 | Mar 2025
p-ISSN: 2395-0072
www.irjet.net
DEEP LEARNING BASED AUTISM BEHAVIOR MONITORING AND EDUCATIONAL REPORT GENERATING SYSTEM Carly Hampson A, Ms. Devipriya M* Department of CSE, School of Computer Science and Engineering, Sathyabama Institute Of Science And Technology, Chennai– 600119, Tamil Nadu, India ---------------------------------------------------------------------***--------------------------------------------------------------------Persistent, stereotyped conduct is a hallmark of autism Abstract - - Integrating a variety of data is necessary for
spectrum sickness (ASD), and early prognosis and intervention can notably enhance the analysis of people with the situation. The faster remedy is obtained, the greater the risk of improvement. However, currently, doctors decide whether or not a patient has ASD based on their conduct and day by day functioning, which is a particularly subjective process. This has created a pressing need for an objective diagnostic approach to help medical doctors make a correct prognosis. With the improvement of present day scientific era and synthetic intelligence, machine learning techniques for reading magnetic resonance imaging (MRI) brain images of ASD sufferers have shown top notch effects. For instance, one examine converted time collection statistics into electricity spectral density for spatial map evaluation, using sparse auto encoders to reduce the size of the enter information into an assist vector system (SVM). Another method is to construct a deep neural network that trains a layered sparse auto encoder to study practical connectivity patterns from a massive problem database.
diagnosing Autism Spectrum Disorder (ASD). Assessments of behaviour, scans of the brain using neuroimaging, and genetic markers. A novel multimodal diagnosis model that is based on Deep Diagram Convolutional Organizations (Deep GCN). Each data type is processed by the model. Separately, locating relevant features, and building a single graph representation that captures intermodal complex relationships. Deep GCN Following that, layers learn hierarchical representations by iteratively aggregating and fusing information to improve the accuracy of diagnostics by utilizing the insights that work together the proposed model uses behavioural, neuroimaging, and genetic data to provide a diagnosis framework for ASD that is both comprehensive and interpretable. Experiments used for validation show that the model works well for integrating multimodal data and enhancing diagnostic capabilities, making available promising headways in clinical choice emotionally supportive networks for chemical imbalance finding. Key Words: Deep Learning, Autism Spectrum Disorder (ASD), Deep Diagram Convolutional Organizations (Deep GCN).
In addition, different studies have used auto encoders to examine purposeful components of whole-brain connectivity and transfer mastering methods to categories large datasets with ASD. Recently, there was a big growth within the collection of non-imaging datasets, which includes affected person genome sequences, gender, and IQ, which play an important role inside the diagnosis of diseases. Combining visualized and non-visualized records using multimodal approaches can improve the overall performance of type algorithms. However, unvisualized facts is often highdimensional, proscribing the competencies of conventional system gaining knowledge of methods. Deep gaining knowledge of gives the opportunity of combining multimodal facts for extra efficient prognosis of mental disorders. For instance, deep mastering can improve mind age and gender estimates, whilst other methods integrate cross-sectional and longitudinal functions estimated using mind MRI. However, non-graphical deep learning strategies are frequently limited to unmarried-pattern programs, which limits their performance. In reaction, graph neural networks (GNNs) had been proposed as a promising solution. Graph convolutional networks (GCNs) expand the remodel feature from Euclidean facts to non-Euclidean graph information to enhance multimodal modelling. One study makes use of GCN to generate graph edges with visible
1.INTRODUCTION People with autism spectrum disorder (ASD) have different social interactions. Behave and communicate. The process of diagnosing ASD involves examining a variety of information, such as genetics, neuroimaging scans, and behavioural assessments data. Integration and interpretation are frequently problematic for traditional methods. Effectively from these diverse sources. Accordingly, this paper proposes a new approach to enhancing ASD that makes use of Deep Graph Convolutional Networks (Deep GCN) diagnosis. Deep GCN is able to separate analyse each type of data and extract key features, and combine them into a single framework that encapsulates intricate relationships between various modalities. By making use of these interconnected experiences, the proposed model intends to work on the precision and comprehension of Diagnosis of ASD This study investigates the potential transformative power of Deep GCN. Clinical procedures by providing a method that is more complete and easier to understand for diagnosing chemical imbalance, eventually planning to further develop results for people influenced by ASD.
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 676