Zepeng "Frazier" Huo

Stanford Center for Academic Medicine · 453 Quarry Rd, Palo Alto, CA 94304 ·

frazierhuo [at] gmail [dot] com

My name is Zepeng Frazier Huo. I'm currently a Postdoc at Stanford University under Shah Lab, conducting research for Foundation Models for medicine. I earned my Ph.D degree from Texas A&M University under STMI Lab, majoring in Computer Science.

My research is towards Knolwdge-based Intelligence for Well-being Informatics (KIWI 🥝). I believe AI should be used for the holistic wellbeing and health of human, i.e. AI for good. The approach to that is to have the responsible AI to learn knowledge instead of blindly absorbing data. This requires many cutting-edge research directions of condensing the knowledge from large-scale data through Foundation Models to learn a strong knowledge base, so that we can achieve the goal more towards autonomous machine intelligence for human wellbeing.

Project Experience

Population-level Pulmonary Embolism Outcome Prediction with Imaging and Clinical Data

National Heart, Lung, and Blood Institute NHLBI: R01HL155410

  • Research student role:
    1. Integrate a comprehensive EMR data set with radiology images to predict Pulmonary Embolism primary and secondary outcomes
  • We published INSPECT dataset at NeurIPS'23 (D/B). Check out our website

Jan 2023 - Present

Warfighter Analytics using Smartphones for Health (WASH)

Defense Advanced Research Projects Agency project: DARPA FA8750-18-2-0027

  • Student managerial leader on TA-2 team:
    1. organize meeting agend
    2. compile monthly report sent to DARPA
  • Research student role:
    1. leading research project on symptom and medical diagnosis prediction using a Mixture-of-Experts (MoE) model on dynamic sensory data;
    2. develop dynamic data imputation for mobile data missingness context;
  • Partial work of this project was presented at AISTATS’20 conference as an oral presentation

June 2019 - Dec 2022

Precise Advanced Technologies and Health Systems for Underserved Population (PATHS-UP)

(National Science Foundation project: FY 2017 1648451, under NSF Engineering Research Center at Texas A&M)

  • Student managerial leader on Thrust-4 team:
    1. organize meeting agend
    2. compile monthly report sent to DARPA
  • Research student role:
    1. Glucose monitoring study for diabetic patients by using a multi-task deep neural network for macro-nutrition prediction from postprandial glucose time-series data;
    2. Design a framework to visualize the glucose response and an interactive UI;
    3. A lead contributor on US patent application ‘PREDICTING FOOD MACRONUTRIENTS FROM BLOOD BIOMARKERS’
  • Partial work of this project was presented at IEEE BHI’19 conference as an oral presentation

Jan 2018 - May 2019

Interpretable patient phenotyping for Emergency department clinical data modeling

(In collaboration with Department of Emergency Medicine at Yale School of Medicine)

  • Responsible for prototyping models and data handling
  • Design a denoising auto-encoder with sparsity for interpretable patient outcome prediction
  • This work was presented at IEEE EMBS’19 conference as a poster

Aug 2018 - May 2019

Contextual activity recognition for wearable daily motion routine capturing

(Project under National Institutes of Health (NIH) of grant 1R01EB028106-01 and National Institute of Biomedical Imagine and Bioengineering, award #1R21EB028486-01)

  • Developing context modeling for understanding daily activity routine for uncertain wearable data
  • Modeling different data types using a multi-modal machine learning framework for pattern recognition under contextual information
  • This work was presented at IEEE BSN’18 conference as a late-breaking poster

Oct 2017 - June 2018


Stanford University, Stanford, CA

Postdoc, Biomedical Informatics
Advisor: Dr. Nigam Shah
Jan 2023 - Present

Texas A&M University, College Station, TX

Ph.D, Computer Science
Advisor: Dr. Bobak Mortazavi
Jan 2018 - Dec 2022

Texas A&M University, College Station, TX

Master of Science, Computer Science
Advisor: Dr. Xia (Ben) Hu
Sep 2015 - Dec 2017

Jilin University, Changchun, China

Bachelor of Engineering, Electrical Engineering
Sep 2011 - Jun 2015


  • ‘Improving Data Efficiency in Contrastive Pre-training for Clinical Multi-modal Models’
    • Zepeng Huo (lead), Alejandro Lozano, Jeya Maria Jose, Mars Huang, Nigam H. Shah, Dev Dash
    • Stanford Institute for Human-Centered Artificial Intelligence (HAI) seed grant
    • Total amount $75,000
    • Duration: 12/1/2023 - 11/30/2024



‘Predicting Food Macronutrients from Blood Biomarkers’

  • United States Patent Application Publication.
  • Publication Number: US 2020/0352481 A1
  • Publication date: Nov. 12, 2020.
  • Predicting a composition of a meal through wearable continuous glucose monitors (CGM) in a multi-task neural network framework.
  • Details can be found in BHI’2019 paper.