您的位置: 首页 >>  学术报告

A Big Data Approach to Understanding Complex Behavioral and Neuroimaging Data

【字体:

Date of event2018-07-11

Time of event: 14:30

LecturerFengqing Zhang

Venue: Rm.303. No.1 Teaching bldg.Yanta Campus

Hosted by: Key Laboratory of Modern Teaching Technology, Ministry of Education, Center for Teacher Professional ability Development

Profile of the Lecturer

Dr. Fengqing Zhang is a tenure-track Associate Professor in the department of Psychology at Drexel University. Her research focuses on the development and application of advanced statistical models to analyze complex and high dimensional data (e.g. neuroimaging data, complex behavioral data). In particular, her lab has been focused on using multimodal neuroimaging (e.g., MRI, DTI, rs-fMRI) to examine neurodegenerative diseases (e.g., Alzheimers disease) and psychiatric disorders (e.g., PTSD, eating disorders). The modeling approach she takes includes machine learning, Bayesian inference, and high dimensional data analysis. In addition, she collaborates with the Weight, Eating, and Lifestyle Science (WELL) center on projects related to treatment development for weight loss maintenance and eating disorders.

As data can be produced and stored more massively and cheaply from various sources, we are entering the era of Big Data. Many traditional statistical models that perform well for moderate sample size do not scale to massive heterogeneous data. Therefore, new statistical thinking and modeling are required.

One important application of big data integration is multimodal neuroimaging. The use of multimodal neuroimaging is a promising and recent approach to study complex brain disorders by utilizing complementary physical and physiological sensitivities. At the same time, however, the advent of multimodal neuroimaging has brought the need to analyze and integrate neuroimaging data with advanced statistical methods that can make full usage of their informational complexity. Using data from the Philadelphia Neurodevelopmental Cohort (PNC) study, we identify three distinct groups, people with trauma exposure and no PTSD symptoms, people with trauma exposure and long-lasting PTSD symptoms as well as healthy controls. A large number of imaging features from different modalities including MRI, DTI, and resting-state fMRI are derived. We then develop an integrative probabilistic model to combine heterogeneous data from multiple modalities and select predictive multimodal imaging signatures of PTSD.

The integrated measurement of diet, physical activity, and the built environment is another important application of big data integration. Recent advances in wearable computing through the use of accelerometers, smartphones, and other devices for tracking individuals and individual behavior, have created a rich opportunity for the integrated measurement of environmental context and behavior. In our weight loss maintenance studies, we develop a smartphone app that utilizes just-in-time adaptive intervention and machine learning to predict and prevent dietary lapses. Data integration strategies using features derived from different sensors to predict affect liability for patients with eating disorders will also be discussed.

版权所有 © 陕西师范大学