Biology subjects

Bhavna, K.

Publications and source records attributed to Bhavna, K..

3 recordsLinked to original sources

bioRxiv2024

Explainable Deep Learning Framework: Decoding Brain Task and Prediction of Individual Performance in False-Belief Task at Early Childhood Stage

Decoding of brain tasks aims to identify individuals brain states and brain fingerprints to predict behavior. Deep learning provides an important platform for analyzing brain signals at different developmental stages to understand brain dynamics. Due to their internal architecture and feature extraction techniques, existing machine learning and deep-learning approaches for fMRI-based brain decoding must improve classification performance and explainability. The existing approaches also focus on something other than the behavioral traits that can tell about individuals variability in behavioral traits. In the current study, we hypothesized that even at the early childhood stage (as early as 3 years), connectivity between brain regions could decode brain tasks and predict behavioural performance in false-belief tasks. To this end, we proposed an explainable deep learning framework to decode brain states (Theory of Mind and Pain states) and predict individual performance on ToM-related false-belief tasks in a developmental dataset. We proposed an explainable spatiotemporal connectivity-based Graph Convolutional Neural Network (Ex-stGCNN) model for decoding brain tasks. Here, we consider a dataset (age range: 3-12 yrs and adults, samples: 155) in which participants were watching a short, soundless animated movie, "Partly Cloudy," that activated Theory-of-Mind (ToM) and pain networks. After scanning, the participants underwent a ToMrelated false-belief task, leading to categorization into the pass, fail, and inconsistent groups based on performance. We trained our proposed model using Static Functional Connectivity (SFC) and Inter-Subject Functional Correlations (ISFC) matrices separately. We observed that the stimulus-driven feature set (ISFC) could capture ToM and Pain brain states more accurately with an average accuracy of 94%, whereas it achieved 85% accuracy using SFC matrices. We also validated our results using five-fold cross-validation and achieved an average accuracy of 92%. Besides this study, we applied the SHAP approach to identify neurobiological brain fingerprints that contributed the most to predictions. We hypothesized that ToM network brain connectivity could predict individual performance on false-belief tasks. We proposed an Explainable Convolutional Variational Auto-Encoder model using functional connectivity (FC) to predict individual performance on false-belief tasks and achieved 90% accuracy.

neuroscience↗

bioRxiv2023

Developmental stability and segregation of Theory of Mind and Pain networks carry distinct temporal signatures during naturalistic viewing

Temporally stable large-scale functional brain connectivity among distributed brain regions is crucial during brain development. Recently, many studies highlighted an association between temporal dynamics during development and their alterations across various time scales. However, systematic characterization of temporal stability patterns of brain networks that represent the bodies and minds of others in children remains unexplored. To address this, we apply an unsupervised approach to reduce high-dimensional dynamic functional connectivity (dFC) features via low-dimensional patterns and characterize temporal stability using quantitative metrics across neurodevelopment. This study characterizes the development of temporal stability of the Theory of Mind (ToM) and Pain networks to address the functional maturation of these networks. The dataset used for this investigation comprised 155 subjects (children (n=122, 3-12 years) and adults (n=33)) watching engaging movie clips while undergoing fMRI data acquisition. The movie clips highlighted cartoon characters and their bodily sensations (often pain) and mental states (beliefs, desires, emotions) of others, activating ToM and Pain network regions of young children. Our findings demonstrate that ToM and pain networks display distinct temporal stability patterns by age 3 years. Finally, the temporal stability and specialization of the two functional networks increase with age and predict ToM behavior.

neuroscience↗

bioRxiv2023

End-to-End Explainable AI: Derived Theory-of-Mind Fingerprints to Distinguish Between Autistic and Typically developing and Social Symptom Severity

Theory-of-Mind (ToM) is an evolving ability that significantly impacts human learning and cognition. Early development of ToM ability allow one to comprehend other peoples aims and ambitions, as well as thinking that differs from ones own. Autism Spectrum Disorder (ASD) is the prevalent pervasive neurodevelopmental disorder in which participants brains appeared to be marked by diffuse variations throughout large-scale brain systems made up of functionally connected but physically separated brain areas that got abnormalities in willed action, self-monitoring and monitoring the intents of others, often known as ToM. Although functional neuroimaging techniques have been widely used to establish the neural correlates implicated in ToM, the specific mechanisms still need to be clarified. The availability of current Big data and Artificial Intelligence (AI) frameworks paves the way for systematically identifying Autistics from typically developing by identifying neural correlates and connectome-based features to generate accurate classifications and predictions of socio-cognitive impairment. In this work, we develop an Ex-AI model that quantifies the common sources of variability in ToM brain regions between typically developing and ASD individuals. Our results identify a feature set on which the classification model can be trained to learn characteristics differences and classify ASD and TD ToM development more distinctly. This approach can also estimate heterogeneity within ASD ToM subtypes and their association with the symptom severity scores based on socio-cognitive impairments. Based on our proposed framework, we obtain an average accuracy of more than 90 % using Explainable ML (Ex-Ml) models and an average of 96 % classification accuracy using Explainable Deep Neural Network (Ex-DNN) models. Our findings identify three important sub-groups within ASD samples based on the key differences and heterogeneity in resting state ToM regions functional connectivity patterns and predictive of mild to severe atypical social cognition and communication deficits through early developmental stages.

neuroscience↗