Skip to content
Tsinghua Alum Xu Hanwen Joins UPenn; His Nature Model Prov-GigaPath Hits 2M Downloads

Tsinghua Alum Xu Hanwen Joins UPenn; His Nature Model Prov-GigaPath Hits 2M Downloads

Tsinghua graduate Xu Hanwen will join UPenn as assistant professor in 2027; his Nature-published Prov-GigaPath model, trained on 1.3 billion pathology images, achieved top performance on 25 of 26 tasks, surpassed 2 million Hugging Face downloads, and entered real-time clinical deployment in 2025.

SmartApe AI
SmartApe AI
SmartApe AI

Tsinghua University alumnus Xu Hanwen, who earned his PhD from the University of Washington under Professor Sheng Wang and held a Weil Family Endowed Fellowship, will join the University of Pennsylvania in spring 2027 as a tenure-track assistant professor in the Departments of Bioengineering and Computer and Information Science. He currently serves as a Visiting Faculty Researcher at Google.

His research centers on the intersection of AI and medicine, focusing on three main areas:

Gigascale Foundation Models for Healthcare

Multimodal Patient Representation Learning

Longitudinal Modeling of Patient Health Trajectories

As first author, Xu published the digital pathology foundation model Prov-GigaPath in Nature . The model was pre-trained on over 1.3 billion pathology images from real-world clinical settings, enabling it to understand whole-slide information from local tissue features. In evaluations covering 26 pathology analysis tasks, Prov-GigaPath achieved the best performance on 25 tasks. After release, the model surpassed 2 million downloads on Hugging Face and was deployed in real-time digital pathology clinical workflows in 2025, providing a new technical pathway for AI-driven cancer analysis.

Building on this work, Xu's team published GigaTIME in Cell (2025), exploring how routine pathology images can infer complex cellular states and spatial molecular information in the tumor microenvironment. This research attracted attention from Microsoft's CEO. In the same year, Pisces appeared in Cell Genomics , fusing eight data modalities to predict synergistic drug combinations.

Xu has also contributed to medical multimodal models such as BiomedCLIP and LLaVA-Rad, with results published in NEJM AI and Nature Communications .

Xu articulates a research philosophy called "a golden loop between AI and medicine": general AI models are tested in real clinical scenarios to discover their capability boundaries, while medical domain expertise and real-world experience feed back to drive AI improvements. This bidirectional cycle aims to bridge foundation model development and medical practice.

His work has been covered by more than 50 global media outlets, including MIT Technology Review, Forbes, and Yahoo.

Original Source

Signed-in readers can open the original source through BestHub's protected redirect.

Sign in to view source
Republication Notice

This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactadmin@besthub.devand we will review it promptly.

multimodal learningAI in healthcarefoundation modelsdigital pathologyGigaTIMEBiomedCLIPPiscesProv-GigaPath
Reader feedback

How this landed with the community

Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.