Generative artificial intelligence-enabled dynamic detection of nicotine-related circuits

TL;DR

This paper introduces a novel generative AI framework that uses dynamic brain network modeling and advanced neural architectures to identify nicotine addiction circuits from functional imaging data, aiding understanding and treatment of addiction.

Contribution

The paper presents an end-to-end generative AI framework combining dynamic brain network modeling, temporal graph Transformer, and contrastive learning for detecting addiction-related circuits from imaging data.

Findings

Successfully models dynamic nicotine addiction circuits

Reveals underlying mechanisms of addiction

Enhances detection accuracy of addiction-related brain networks

Abstract

The identification of addiction-related circuits is critical for explaining addiction processes and developing addiction treatments. And models of functional addiction circuits developed from functional imaging are an effective tool for discovering and verifying addiction circuits. However, analyzing functional imaging data of addiction and detecting functional addiction circuits still have challenges. We have developed a data-driven and end-to-end generative artificial intelligence(AI) framework to address these difficulties. The framework integrates dynamic brain network modeling and novel network architecture networks architecture, including temporal graph Transformer and contrastive learning modules. A complete workflow is formed by our generative AI framework: the functional imaging data, from neurobiological experiments, and computational modeling, to end-to-end neural networks, is transformed into dynamic nicotine addiction-related circuits. It enables the detection of addiction-related brain circuits with dynamic properties and reveals the underlying mechanisms of addiction.