Improved Modeling of Multi-Sensor Mobile Health Data
Wednesday, November 13, 2019
2:30 PM-3:30 PM
The department of Epidemiology and Biostatistics at Dornsife presents Zoe Zhang, PhD, assistant professor in the department of Psychology at Drexel University.
Technology advancements in smartphones and wearable sensors have greatly increased our capability of monitoring, tracking, and transmitting health and behavioral data continuously and in real time. At the same time, it requires the development of new statistical methods that can make full usage of the informational complexity. In our weight loss maintenance and eating disorders studies, we develop a smartphone app that utilizes just-in-time adaptive intervention and machine learning to predict and prevent dietary lapses. Statistical modeling challenges including imbalanced data and missing data will be discussed. A cost-sensitive ensemble model is proposed as a meta-technique to combine multiple weak classifiers and introduce cost items into the learning framework. Additionally, we develop a new data integration strategy based on multiple kernel learning to combine features from different sensors to predict emotional eating episodes. A large number of physiological features are extracted from different wearable sensors in both time and frequency domains. Analysis results show that our proposed approach is efficient and effective for real time prediction.
Zhang is a tenure-track assistant professor in the department of Psychology at Drexel University. Her research interests lie primarily in the development and application of advanced statistical models to analyze complex and high dimensional data (e.g., neuroimaging data, complex behavioral data). In particular, she has focused on neuroimaging statistics, data mining, and high dimensional data analysis. Through interdisciplinary collaborations, she has extended her statistical expertise to several applied areas, including neuroimaging, wearable computing, weight loss maintenance, and eating disorders.