Sign In

Wei-Hung Weng

PhD Candidate @ MIT CSAIL

Wei-Hung Weng is a multidisciplinary researcher with expertise in artificial intelligence, machine learning, and healthcare. He completed his PhD in Computer Science and Artificial Intelligence at MIT CSAIL from 2017 to 2022.14 Prior to his doctoral studies, Weng obtained an MD from Chang Gung University in Taiwan and an MMSc in Biomedical Informatics from Harvard Medical School.14

Weng's professional journey includes:

  1. Currently working as a Research Scientist at Google since February 2022.4
  2. Served as a PhD Research Assistant at MIT CSAIL from September 2017 to May 2022.4
  3. Worked as a Consultant for Vysioneer from January 2019 to December 2021.4
  4. Held positions as a Postdoctoral Research Fellow at Massachusetts General Hospital and a Research Associate at National Taiwan University.4
  5. Practiced as a Resident Doctor at Chang Gung Memorial Hospital in Taiwan.4

His research focuses on applying machine learning algorithms to healthcare, particularly in areas such as multimodal representation learning, reinforcement learning, and clinical decision making.1 Weng has contributed to numerous publications in the fields of artificial intelligence, machine learning, and healthcare informatics.23

Highlights

Apr 30 · linkedin

Glad to introduce Med-Gemini, a family of Google Gemini-based models specialized for medicine! https://lnkd.in/g7KVExAV Our key contributions include:

  • Advancing clinical reasoning capabilities via self-training with web search, and uncertainty-guided search at inference
  • Improving multimodal understanding with fine-tuning
  • Leveraging long-context capabilities, the core strength of Gemini, with chain-of-reasoning
  • Achieving state-of-the-art performance across various challenging medical benchmarks, including MedQA-USMLE and NEJM CPC diagnosis prediction, multimodal tasks such as NEJM image challenges, and several long-context problems in medical video and long EHR
  • Demonstrating promising real-world utility on tasks: medical text summarization, referral letter generation, QA on multiple biomedical papers, and multimodal medical dialogue

Notably, Med-Gemini’s long-context capability allows it to process and understand lengthy (1M+ context tokens) and complex medical data, such as EHRs and videos, opening up previously infeasible medical application possibilities—we have achieved remarkable performance on a needle-in-a-haystack retrieval task from very long de-identified health records, medical instruction video question answering, surgical video action annotation, and surgical video view assessment.

Furthermore, we showcase the great potential of Med-Gemini being an articulate conversational agent in multimodal medical dialogue, handling radiology / dermatology images, surgical videos, long EHRs, and even a bundle of biomedical research papers!

We envision a promising future where such innovative and capable medical AI systems can safely accelerate scientific progress and improve medical care.

Privileged to work with talented people Google Research Google DeepMind Khaled Saab Tao Tu Ryutaro Tanno Vivek Natarajan Alan Karthikesalingam MD PhD Juraj Gottweis S. Sara Mahdavi David Stutz Ellery Wulczyn Fan Zhang Tim Strother Chunjong Park Elahe Vedadi Juanma Zambrano Chaves Szu-Yeu(Jimmy) Hu Mike Schäkermann Aishwarya Kamath Yong Cheng David Barrett Cathy Cheung Basil Mustafa Anil Palepu Daniel McDuff Le Hou Tomer Golany Luyang Liu Jean-Baptiste ALAYRAC Neil Houlsby Nenad Tomašev Jan Freyberg Jonathan Krause Christopher Semturs Dale Webster Greg Corrado Yossi Matias Ewa Dominowska S. M. Ali Eslami Jonathon Shlens Claire Cui Oriol Vinyals Joëlle Barral koray kavukcuoglu James Manyika Jeff Dean Demis Hassabis

Related Questions

What are some of Wei-Hung Weng's most notable research projects?
How did Wei-Hung Weng transition from being a physician to a researcher in AI?
What are the main applications of machine learning in healthcare according to Wei-Hung Weng?
Can you provide examples of Wei-Hung Weng's contributions to multimodal representation learning?
What are the key challenges Wei-Hung Weng has faced in his research on clinical decision making?
Wei-Hung Weng
Add to my network

Location

Cambridge, Massachusetts, United States