Zhicheng (Stark) Guo

prof_pic.jpg

I’m a postdoctoral researcher at Duke University working with Professor Cynthia Rudin. I received my Ph.D. in Electrical and Computer Engineering from Duke in 2026 and my B.Sc. in Computer Science from Rensselaer Polytechnic Institute.

I develop interpretable machine learning methods for scientific discovery and clinical decision making, with a focus on foundation models and unsupervised concept discovery. My goal is to turn complex learned representations into reusable, human-understandable concepts that help experts uncover meaningful knowledge in data.

My recent work focuses on unsupervised concept discovery from foundation models: identifying reusable, human-understandable concepts in learned representations without manual concept annotations. I aim to connect the expressive power of foundation models with domain knowledge, enabling experts to examine model reasoning and uncover meaningful patterns in complex data.

More broadly, my work spans interpretable deep learning, multimodal representation learning, and human–AI collaboration, with applications in medical imaging and physiological monitoring. I have developed methods that uncover shared structure across sensing modalities, reveal clinically meaningful relationships in data, and support expert decision making. In a user study, my interpretable EEG model improved medical professionals’ classification accuracy from 47% to 71%.

My work has appeared in NEJM AI, Nature Machine Intelligence, JMLR, CVPR, and Harvard Data Science Review.

selected publications

View all
  1. ProtoCDisco: Unsupervised Concept Discovery Using Foundation Model Representations Manuscript in submission, 2026
  2. A Mammographic Texture Lexicon for Interpretable Breast Cancer Risk Prediction Ongoing research, 2026 · Work in progress
  3. Improving Clinician Performance in Classifying EEG Patterns on the Ictal--Interictal Injury Continuum Using Interpretable Machine Learning NEJM AI, 2024
  4. Improving Atrial Fibrillation Detection Using a Shared Latent Space for ECG and PPG Signals Harvard Data Science Review, 2025

Research map

How the papers connect

Each node is a paper. Select a topic or paper to trace the shared methods and application areas across my research.

An interactive network of publications grouped into wearable cardiac monitoring, AI for Scientific Discovery and Clinical Decision Aid, and Interpretable AI.

news

Jul 01, 2026 I started a new role as a postdoctoral researcher at Duke University, developing foundation models and interpretable AI for medical imaging.
May 01, 2026 I completed my Ph.D. in Electrical and Computer Engineering at Duke University!
Aug 30, 2025 Wrapped up my machine learning research internship at Texas Instruments Kilby Labs!
May 30, 2025 I will attend CVPR 2025 in Nashville. See you there!
May 14, 2025 I started my summer internship as a machine learning researcher at Kilby Labs, Texas Instruments, in Dallas, Texas. I will be working on LLMs and embedded AI.