Machine Learning Approaches in Predicting ASD Using Gut Microbiome Profile

The most reliable method of predicting Autism Spectrum Disorder (ASD) is using machine learning algorithms and the gut microbial profile has provided a promising advancement in neurodevelopment conditions. Recently World Health Organization (WHO) has lighted on the significant role of gut microbiome profile in neurodevelopmental conditions. This metagenomics study uses the aid of 16s rRNA sequencing technology to analyse the gut microbiome profile which plays a pivotal role in neurodevelopmental conditions like ASD. The proposed approach employed various Machine Learning Tools like J48, Artificial Neural Network (ANN), Random Forest (RF) and Naïve Bayes to predict and classify ASD, which were implemented in the Weka software. These models were alone chosen due to their high-performance accuracy. Additionally, the proposed work utilized Omics Box for quality control and to analyse the microbiome abundance level. The proposed work uses Machine learning (ML) to study the gut microbiome and its role to the yearly diagnosis of ASD. This work applies J48, Random Forest(RF), Artificial Neural Network(ANNs) and Naive Bayes(NB). ML classifiers are used to classify and obtain models to predict ASD and Neurotypical(NT) based on the gut microbial composition. This approach was analysed with two ASD datasets and the significant classification outcome compared to traditional tests and data-based approaches. The application of ML yields a classification accuracy as high as 98%, with a sensitivity of 99% and specificity reaching 98%. Our findings suggest that the overabundance of genus Bacteroides, Clostridium XLVa and Prevotella has prevailed in ASD conditions. This research integrates the ML algorithms and Gut microbiome profile analysis, which strengthens the early detection and understanding of ASD. This paves for a novel approach to the diagnostic and therapeutic strategies.