I-Corps: Utilizing Machine learning and Artificial Intelligence (AI) for Early Detection and Identification of Mental Disorders

点击翻译按钮获取中文摘要

The broader impact of this I-Corps project is the development of automated diagnosis tools for neurological disorders. Currently, there is no single quantitative test (like a blood glucose test) that can be done to diagnose neurological disorders. The proposed technology is based on advanced deep learning models applied to multi-modal imaging, such as magnetic resonance imaging (MRI) and electroencephalography (EEG), for early detection. The technology may be used to distinguish abnormal brain scans from healthy scans and may be able to aid in the diagnosis of neurological disorders such as autism spectrum disorder (ASD), attention deficit hyperactivity disorder (ADHD), Alzheimer’s disease (AD), epilepsy, and Parkinson’s disease (PD). For example, current ADHD diagnostic tests result in the mis-, over-, or under-diagnosis of nearly a million children annually. If successful, the proposed technology may lead to enhanced diagnostic accuracy and improved clinical outcomes. This I-Corps project is based on the development of advanced machine learning algorithms including deep learning (DL), network science principles, and data-driven approaches that may identify markers for neurological disorders. The goal is to provide a diagnostic test that is non-invasive, represents progression of the neurological disorder, correlates with symptomatology, and provides early-detection. A single quantitative test does not exist for most neurological disorders or relies on single-modality data for disorders such as epilepsy. Preliminary data suggests that applying DL methods to multiple modalities of brain data (functional MRI - fMRI, structural MRI - sMRI, EEG) will lead to a transformative artificial intelligence (AI)-based, quantitative strategy to diagnose these disorders – without the limitation of the current clinical methods. Successful models for the diagnosis of ASD and ADHD have been designed and results indicate an accuracy improvement of up to 28% using MRI data alone. The technology for early detection of epilepsy and AD are currently under development. These models will serve as proof-of-concept and a base for developing solutions for multi-disorder diagnosis and prediction using multi-modal imaging data.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.