Using connectome-based predictive modeling to predict individual behavior from brain connectivity

• Published in: Nature protocols Vol. 12; no. 3; pp. 506 - 518
• Main Authors: Shen, Xilin, Finn, Emily S, Scheinost, Dustin, Rosenberg, Monica D, Chun, Marvin M, Papademetris, Xenophon, Constable, R Todd
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 01.03.2017; Nature Publishing Group
• Subjects: 59/36; 59/57; 631/114/794; 631/1647/2198; 631/1647/245; 631/378/116/2396; Analytical Chemistry; Behavior; Biological Techniques; Brain; Brain - anatomy & histology; Brain - cytology; Brain - physiology; Brain mapping; Brain research; Computational Biology/Bioinformatics; Connectome - methods; EEG; Electroencephalography; Feature selection; Functional magnetic resonance imaging; Humans; Innovations; Learning algorithms; Life Sciences; Machine learning; Magnetic resonance imaging; Medical imaging; Medicine; Microarrays; Modelling; Models, Neurological; Nerve Net - anatomy & histology; Nerve Net - cytology; Nerve Net - physiology; Neural circuitry; Neural networks; Neural Pathways; Neuroimaging; Neurosciences; Optimization; Organic Chemistry; Permutations; Prediction models; Predictions; Protocol; Psychological aspects; Tensors; United States > US; Connecticut
• ISSN: 1754-2189; 1750-2799; 1750-2799
• DOI: 10.1038/nprot.2016.178

This protocol describes how to develop linear models to predict individual behavior from brain connectivity data with proper cross-validation, and how to use an online tool to visualize the most predictive features of the models. Neuroimaging is a fast-developing research area in which anatomical and functional images of human brains are collected using techniques such as functional magnetic resonance imaging (fMRI), diffusion tensor imaging (DTI), and electroencephalography (EEG). Technical advances and large-scale data sets have allowed for the development of models capable of predicting individual differences in traits and behavior using brain connectivity measures derived from neuroimaging data. Here, we present connectome-based predictive modeling (CPM), a data-driven protocol for developing predictive models of brain–behavior relationships from connectivity data using cross-validation. This protocol includes the following steps: (i) feature selection, (ii) feature summarization, (iii) model building, and (iv) assessment of prediction significance. We also include suggestions for visualizing the most predictive features (i.e., brain connections). The final result should be a generalizable model that takes brain connectivity data as input and generates predictions of behavioral measures in novel subjects, accounting for a considerable amount of the variance in these measures. It has been demonstrated that the CPM protocol performs as well as or better than many of the existing approaches in brain–behavior prediction. As CPM focuses on linear modeling and a purely data-driven approach, neuroscientists with limited or no experience in machine learning or optimization will find it easy to implement these protocols. Depending on the volume of data to be processed, the protocol can take 10–100 min for model building, 1–48 h for permutation testing, and 10–20 min for visualization of results.