A Bayesian Framework for Joint Analysis of Heterogeneous Neuroscience Data

Project Summary

This paper addresses analysis of heterogeneous data, such as ordered, categorical, real and count data. Such data are of interest in our motivating application, cognitive and brain science, in which subjects may answer questionnaires, and also (separately) undergo fMRI interrogation. A contribution of this paper concerns the joint analysis of how people answer questionnaires and how their brain responds to external stimuli (here visual), the latter measured via fMRI.

Themes and Categories
Year

In this paper we ask a novel and practical question, which to our knowledge has not been considered previously: can one predict the fMRI response (here from the amygdala and the ventral striatum) to external stimuli, based upon knowledge of how the subject answers a questionnaire and genetic data?

A new model is developed for joint analysis of ordered, categorical, real and count data. In the motivating application, the ordered and categorical data are answers to questionnaires, the (word) count data correspond to the text questions from the questionnaires, and the real data correspond to fMRI responses for each subject. We also combine the analysis of these data with single-nucleotide polymorphism (SNP) data from each individual. The questionnaires considered here correspond to standard psychological surveys, and the study is motivated by psychology and neuroscience. The proposed Bayesian model infers sparse graphical models (networks) jointly across people, questions, fMRI stimuli and brain activity, integrated within a new matrix factorization based on latent binary features. We demonstrate how the learned model may take fMRI and SNP data from a subject as inputs, and predict (impute) how the individual would answer a psychological questionnaire; going in the other direction, we also use an individual's SNP data and answers from questionnaires to impute unobserved fMRI data. Each of these two imputation settings has practical and theoretical applications for understanding human behavior and mental health, which are discussed.

http://people.ee.duke.edu/~lcarin/Esther_ICML_finalv2.pdf

http://people.ee.duke.edu/~lcarin/Esther_JASA_7.pdf

 

 

Related People

Related Projects

A team of students led by Professor Anru Zhang (Duke Biostatistics & Bioinformatics, Computer Science, Mathematics, and Statistical Science) will develop methods to investigate the courses of complex diseases through electronic health records. The team will apply tensor methods to identify key features to register the patient's timeline. This work will provide a basis for researchers at Duke and elsewhere to sufficiently utilize the high-dimensional and longitudinal information in the electronic health record data. If time permitted, the related methods and theory will also be studied.

 

Project Lead: Anru Zhang

Project Manager: Rungang Han

A team of students led by researchers in the BIG IDEAs Lab will work to create a cloud-based infection detection platform that populates and translates wearable data from a variety of sources. The project will involve working with existing wearable data pipelines (e.g., APIs) to collect, process, and visualize wearable device data in real-time as well as implement machine learning techniques to produce insights of user health. The project will also involve app development and UI/UX considerations so that the platform is easily accessible and user friendly. The ultimate goal of this work is to inform wearable device users of changes in their health condition before more serious symptoms occur. 

 

Project Lead: Ali Roghanizad

A team of students led by researchers at the Duke Center for Policy Impact in Global Health (CPIGH) will create a user-friendly interactive visualization tool to track the evolution of Universal Health Coverage (UHC) financing policies in the low- and middle-income countries. The students will use the UHC policy surveillance data collected by the CPIGH Bass Connections team in the Fall 2021 and Spring 2022 semesters. This publicly available data visualization will provide an opportunity for policymakers and researchers to conduct a quick comparative cross-country analysis and longitudinal analysis to understand the impact of different policy experiments on UHC progress in different jurisdictions.

Project Lead: Osondu Ogbuoji

Project Manager: Luke Fesko