About Me
I’m a first year PhD student in the Human Signals Lab at UT Austin ECE, supervised by Prof. Edison Thomaz. I received B.S. Electrical Engineering and B.A. Statistics degrees from Rice University. My research interests include Ubiquitous and Mobile Computing, Mobile Health, Wearable Computing, Activity Recognition, Machine Learning and Signal Processing. I want to build systems and models to understand people’s behaviors and improve their health and quality of life. During my undergraduate study, I had research experience in the Computational Wellbeing Group under the supervision of Prof. Akane Sano, where I worked on machine learning models for wellbeing predictions with mobile and wearable sensing. I also worked with Prof. Behnaam Aazhang on Alzheimer’s Disease diagnosis with MRI and machine learning. Here is my CV for more information.
Education
Ph.D. Electrical Engineering
Supervisor: Prof. Edison Thomaz
B.S. Electrical Engineering, B.A. Statistics
Magna Cum Laude
Research Experience
Sleep Advice Prediction with Multimodal Data
Supervisor: Prof. Akane Sano, Rice University
- Developed user-dependent and user-independent Random Forest models to predict three doctors’ sleep advice to shift workers in two Japanese hospitals using daily survey and wearable device data
- Output advice probabilitites to indicate models’ confidence in advice selections
- Compared the performance among individual models built for each doctor and a one-size-fits-all model
- Delivered the models to collaborators in Japan for a clinical study to evaluate the effectiveness of sleep advice to shift workers
Schizophrenia Patients’ Symptom Prediction with Mobile Phone Sensing
Supervisor: Prof. Akane Sano, Rice University
- Developed user-dependent and user-independent Gaussian Process and Long Short-Term Memory regression models to predict Schizophrenia patients’ self-assessment scores of ten symptoms with behavioral and environmental data collected from mobile phone sensing
- Personalized the independent models by fine-tuning transfer learning and clustering participants’ daily behavior patterns and demographic information
- Extended the system from predicting symptoms for only one day to thirty days with a Long Short-Term Memory classification model which would give doctors more time to prepare interventions for patients in practical use
Alzheimer’s Disease Diagnosis with Self-supervised Learning and MRI
Supervisor: Prof. Behnaam Aazhang, Rice University
- Developing classification models for subjects’ cognitive conditions and Alzheimer’s Disease diagnosis with Magnetic Resonance Imaging
- Working on Convolutional Neural Network models with Self-supervised Learning to demonstrate the possibility of accurate disease diagnosis with only a limited amount of labeled data
- Gathered benchmark MRI datasets for pre-training and classification tasks