Research

Artificial intelligence and sensor-based autism spectrum disorder diagnosis

Autism spectrum disorder (ASD) is a complicated neurodevelopmental condition whose cause is unclear, and due to its unusual pattern, it is difficult to diagnose the disease at the right time. Because of its ability to automatically uncover complicated patterns in high-dimensional data, artificial intelligence (AI) can be a beneficial tool. Recent improvements in neuroimaging technologies using biosensors have enabled the quantification of functional abnormalities associated with ASD. This work proposes an approach to constructing a functional connectivity network from resting state functional Magnetic Resonance Image (rs-fMRI) data. The time series part of fMRI data is used to make a functional connectivity network. From this, a correlation matrix is made that shows how much the different parts of the brain interact with each other. Several brain atlases have been considered in the experiment. With the majority voting concept based on the results from the atlas, the proposed technique reveals an ASD detection accuracy of 84.79%.

AIM-Scan: AI-enabled Mobile Healthcare Unit to Detect the Emotional State of Autistic Children

Artificial Intelligence can play an important role in smart monitoring and supportive housing facilities for patients suffering from neuro-degenerative disorders. Autistic children have challenges with social skills, recurrent activities, verbal and nonverbal communication, and adapting to their surroundings. As a result, dealing with autistic children is a severe public health concern since it is difficult to tell what they are experiencing due to a lack of emotional and cognitive skills. Unfortunately, to date, no proper treatment has been discovered for autism and it is considered an incurable disease. This study focuses on developing cognitive capacity and emotional quotient, as well as enhancing the autistic child’s ability to function and engage constructively in society. To solve this problem we have considered Artificial Intelligence and Internet of Things devices such as mobile and headsets to build a friendly and assistive environment for autistic children.

EEG-based Emotion Prediction based on Audio Visual Stimulation using En-FS, XG-Boost, and SEn-G model

In this paper, we have thoroughly studied the data and experimented with various techniques to find out the best one and according to that we have designed the model to classify a maximum number of emotions in a short time. We have extracted various domains of features including time, frequency, wavelet, etc., and applied the Ensemble Feature Selection method to find out the best subset of features. This technique includes correlation technique, information gain measurement, and finally recursive feature elimination method to find the optimized feature set. For the classification part, various method has been reviewed and then a stacked ensemble generalization model has been adopted with respect to bagging and boosting results of state-of-the-art machine learning techniques. The results show that the ensemble approach of feature optimization (En-FS) combined with the stacked ensemble generalized model (SEn-G) classification performs better. The suggested technique has been tested using the DEAP Dataset, and the experimental findings support the efficacy of the strategy while also being compared to state-of-the-art approaches.

SEC-EnD: Stacked Ensemble Correlation-based Feature Selection Method for Emotion Detection

In this research, we provide SEC-END, a stacked ensemble correlation-based feature selection method for emotion detection that can combine and stack different feature selection methods to find the best features. This method has a 3-level feature selection technique, which helps to fine-tune the selection of features. After integrating the entire feature set, we took the features that each approach had in common and used the union quorum method to eliminate the unnecessary features. Then, using four classifiers-Random Forest, Decision Tree, SVM, and KNN—we tested the feature set. Regardless of the training set and classifier, the feature subset is outperforming other state-of-the-art techniques for all the dimensions. The experiment is also carried out with different sizes of windows for EEG signals so that the optimum window size can be recognized and used for further experiments.

A Fusion Feature Technique for F-Connectivity Analysis of Brain to Recognize Human Emotion

This research aims to develop an algorithm for accurately detecting human emotions from Electroencephalogram (EEG) signals. The study focuses on comparing the effectiveness of the brain connectivity features, namely Phase Slope Index (PSI), and Phase Lag Value (PLV), and Empirical Mode Decomposition (EMD) domain features in classifying emotions. A multidomain feature subset is proposed by fusing connectivity domain features with features from the Empirical Mode Domain. To improve the classification performance a hybrid F-score based SVM feature selection has been implemented to identify the most discriminative brain connectivity feature subset among all extracted features. Further, advanced correlation-based feature selection has been developed for EMD domain feature selection. Furthermore, a Multilevel Stacked Ensemble Mode classifier is proposed to handle features from both domains and achieve maximum accuracy in emotion classification. The outcomes of this research contribute to the development of computational models for emotion detection using EEG signals, with potential applications in affective computing.

An Improved Cognitive Approach for Automated Epileptic Seizure Detection from Multichannel EEG

Recently, various automated seizure detection frameworks based on machine learning techniques have been suggested to replace traditional approaches. But as multichannel EEG data is chaotic, choosing optimum channels as well as features and categorizing them are still open challenges. In this paper, we have proposed one method called sequential channel optimization by which we will select the optimum channels or electrodes to classify the data. After that, we have extracted time domain, frequency domain, wavelet domain, and EMD domain features and fed the data through SVM, KNN, and Random Forest classifiers. It has been noticed that after channel optimization the accuracy has been improved and EMD features outperform all the other domains to detect epilepsy. It is also noticeable that SVM performs best with EMD features.

A Study on Seizure Detection Performance in an Automated Process by Extracting Entropy Features

Millions of people of all ages have been diagnosed with epilepsy all over the world. Electroencephalography (EEG), a quantitative component, is vital in identifying and analyzing epileptic seizures. Manual EEG detection takes a long time and has serious consequences. To prevent this circumstance, the world needs alternative detection technologies. For more than a decade, several strategies and procedures have been used to assist medical doctors in detecting technological improvement. Numerous automated seizure identification frameworks that use machine learning approaches have recently been presented to replace traditional methodologies. However, because multichannel EEG data is unpredictable, selecting optimal channels as well as characteristics and classifying them remain unsolved issues. During automatic signaling, the gadget also emits a noisy signal, making detection and prediction challenging. In this paper, we have tested various entropy features and selected the best features to classify seizure patients. First, we preprocessed the data using IMF of the raw signal, and then entropy features were extracted to classify the epileptic patients. The proposed channel selection method, SVM-GA, works well with our framework and successfully removes redundant channels from the data

Page updated

Google Sites

Report abuse