TY - GEN
T1 - Visualizing Functional Network Connectivity Difference between Healthy Control and Major Depressive Disorder Using an Explainable Machine-learning Method
AU - Chun, Ji Ye
AU - Sendi, Mohammad S.E.
AU - Sui, Jing
AU - Zhi, Dongmei
AU - Calhoun, Vince D.
N1 - Publisher Copyright:
© 2020 IEEE.
PY - 2020/7
Y1 - 2020/7
N2 - Major depressive disorder (MDD) is a complex mental disorder characterized by a persistent sad feeling and depressed mood. Recent studies reported differences between healthy control (HC) and MDD by looking to brain networks including default mode and cognitive control networks. More recently there has been interest in studying the brain using advanced machine learning-based classification approaches. However, interpreting the model used in the classification between MDD and HC has not been explored yet. In the current study, we classified MDD from HC by estimating whole-brain connectivity using several classification methods including support vector machine, random forest, XGBoost, and convolutional neural network. In addition, we leveraged the SHapley Additive exPlanations (SHAP) approach as a feature learning method to model the difference between these two groups. We found a consistent result among all classification method in regard of the classification accuracy and feature learning. Also, we highlighted the role of other brain networks particularly visual and sensory motor network in the classification between MDD and HC subjects.
AB - Major depressive disorder (MDD) is a complex mental disorder characterized by a persistent sad feeling and depressed mood. Recent studies reported differences between healthy control (HC) and MDD by looking to brain networks including default mode and cognitive control networks. More recently there has been interest in studying the brain using advanced machine learning-based classification approaches. However, interpreting the model used in the classification between MDD and HC has not been explored yet. In the current study, we classified MDD from HC by estimating whole-brain connectivity using several classification methods including support vector machine, random forest, XGBoost, and convolutional neural network. In addition, we leveraged the SHapley Additive exPlanations (SHAP) approach as a feature learning method to model the difference between these two groups. We found a consistent result among all classification method in regard of the classification accuracy and feature learning. Also, we highlighted the role of other brain networks particularly visual and sensory motor network in the classification between MDD and HC subjects.
UR - http://www.scopus.com/inward/record.url?scp=85091002036&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85091002036&partnerID=8YFLogxK
U2 - 10.1109/EMBC44109.2020.9175685
DO - 10.1109/EMBC44109.2020.9175685
M3 - Conference contribution
C2 - 33018257
AN - SCOPUS:85091002036
T3 - Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS
SP - 1424
EP - 1427
BT - 42nd Annual International Conferences of the IEEE Engineering in Medicine and Biology Society
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 42nd Annual International Conferences of the IEEE Engineering in Medicine and Biology Society, EMBC 2020
Y2 - 20 July 2020 through 24 July 2020
ER -