Yuan Luo, PhD

Current Affiliation: Assistant Professor, Department of Preventive Medicine, Division of Health and Biomedical Informatics, Feinberg School of Medicine, Northwestern University
Education:
PhD, Computer Science, MIT
MS, Computer Science, State University of New York at Albany
BS, Electrical Engineering, Tsinghua University, Beijing

How do I describe my work to those outside the field ...

My research applies machine learning and data sciences to automate the transformation from cases and experiences to medical knowledge. My group aims to integrate multiple types of data, — clinical notes, structured Electronic Health Record (EHR) data, images, and -omic data — to facilitate evidence based, personalized therapies and to inform clinical trials. I had the unique opportunity to reach out and work closely with researchers with diverse backgrounds. I work with hematopathologists and oncologists at Massachusetts General Hospital and Northwestern Memorial Hospital on applying machine learning to understand cancer pathology. I work with physicians and geneticists at Boston Children’s Hospital and Northwestern Memorial Hospital on adapting a network-wide association study to understand the genetic predisposition of hereditary mental and cardiovascular diseases. I work with clinical pathologists at Massachusetts General Hospital on quantifying information redundancies and potential diagnostic synergies within sets of common laboratory tests. Wide collaborations benefited me with perspectives from different biomedical subdomains, and motivated me to develop methods that have both better accuracy and better interpretability over state-of-the-art models. 

Years of experience:

I’ve been in informatics for 12 years.

Why Informatics?

Throughout my undergrad back in China, I took many courses in biology and chemistry. Together with my major coursework in electrical engineering, they built my foundations for working in the biomedical informatics field. Then I came across my PhD co-advisors, Dr. Ozlem Uzuner and Dr. Peter Szolovits, whose research immediately drew my interest at a time when I was prepared. 

What are your ambitions? At the end of your career, what do you hope to have accomplished?

The goal is pretty clear and is in fact the mission of my group: to integrate multiple modalities of data ranging from clinical narrative text to structured EHR data to images and to -omic data in order to inform targeted therapy and clinical trials. At the end of my career, I hope to have built the suite of machine learning and analytics tools to realize the goal, and to have showcased the benefit of such realization in transforming and improving patient care.

Who or what are your “key sources” in the informatics field?

My key sources include both top informatics journals such as JAMIA, JBI, Bioinformatics, Briefings in Bioinformatics etc. as well as top machine learning and AI conferences such as AAAI, ICML etc. I also like going to the AMIA meetings to listen to the talks and exchange ideas with colleagues. I believe that the diversity of key sources helps to build the diversity of skill set.

Articles that spotlight my research interest ...

  1. Y Luo*, C Mao, Y Yang, F Wang, FS Ahmad, D Arnett, MR Irvin, SJ Shah. Integrating Hypertension Phenotype and Genotype with Hybrid Non-negative Matrix Factorization. Bioinformatics: 2018 Sep 15. doi: 10.1093/bioinformatics/bty804. PMID: 30239588
  2. Y Luo, Y Cheng, O Uzuner, P Szolovits, J Starren. Segment convolutional neural networks (Seg-CNNs) for classifying relations in clinical notes. Journal of the American Medical Informatics Association (JAMIA) 2018 25(1) 93-98, doi: 10.1093/jamia/ocx090 PMID: 29025149.
  3. Y Luo, Y Xin, R Joshi, L Celi, P Szolovits. Predicting ICU Mortality Risk by Grouping Temporal Trends from a Multivariate Panel of Physiologic Measurements. Proceedings of AAAI Conference on Artificial Intelligence (AAAI, acceptance rate 26%) 2016 (pp. 42-50).
  4. Y Luo, Y Xin, E Hochberg, R Joshi, O Uzuner, P Szolovits. Subgraph Augmented Non-Negative Tensor Factorization (SANTF) for Modeling Clinical Text. Journal of the American Medical Informatics Association (JAMIA) 2015 22(5) 1009-1019 doi: 10.1093/jamia/ocv016. PMCID: 4986663

Hobbies/Interests outside AMIA ...

I did competitive ballroom dancing and I enjoy hiking.

AMIA is important to me because ...

I consider AMIA as my professional home where I connect with informatics colleagues. AMIA brings together researchers and professionals from diverse backgrounds all over the world. This constantly broadens my horizons and shapes my way of thinking through the healthcare field.

I am involved with AMIA ...

I am on the editorial board for JAMIA Open and PLOS One (medical informatics area) and am a member of the AMIA Membership and Outreach Committee. I have been a long-time member of the AMIA Natural Language Processing Working Group (NLP-WG) and Knowledge Discover and Data Mining Working Group (KDDM-WG). I have also served on the Scientific Program Committee for multiple AMIA Annual Symposiums and AMIA Informatics Summits meetings.

It may surprise people to know ...

I did competitive ballroom dancing. I used to compete for the MIT Ballroom Dance Team, and enjoyed a different way of thinking of life through dancing. Eventually I quit competitive ballroom dancing to devote more time to AMIA (tongue-in-cheek).