Uri Kartoun


Uri Kartoun


Staff Research Scientist & IBM Master Inventor


IBM Research - Cambridge Cambridge, MA USA


  • Collaborating with Massachusetts General Hospital I co-created MELD-Plus—a highly accurate risk score in liver disease. The score received the attention of Transplantation in June 2020 and Journal of Hepatology in July 2022.
  • In March 2022 I helped releasing an R packge containing a new type of a feature selection method. I co-invented and developed the method with collaborators from MIT—the method seems to be working well (tested on multiple databases includeing the UK Biobank) and may serve as a possible substitute for widely used methods such as Random Forests and adaptive Lasso. 
  • I have co-invented more than 50 inventions and was appointed IBM Master Inventor in Oct. 2019.
  • I created EMRBots helping thousands to practice machine-learning algorithms, publish papers, and advance teaching by using simulated electronic medical records. EMRBots were used to publish more than 20 manuscripts by other researchers (including in KDD / IEEE conferences and journals such as Bioinformatics).
  • In 2008 I went through an emergency 3-hour lung surgery and this experience inspired me to identify possible opportunities to improve outcomes—I published that in the European Respiratory Journal and in the Annals of Surgery with a hope to affect future clinical guidelines in lung. It may sound surprising, but this experience helped me also to identify an opportunity to improve performance of prediction mechanisms; clinicians' intuition may be measured and incorporated into machine learning algorithms to improve performance—I published the concept in the Journal of Medical Systems with a hope that the new intuition covariates will indeed improve performance, maybe even significantly.
  • Collaborating with Merck & Co. & Harvard, I co-developed the first algorithm to classify insomnia patients.
  • I joined IBM Research in Sept. 2016.
  • I was a research fellow at Harvard Medical School/Massachusetts General Hospital (2013–2016).
  • I was a software development engineer at Microsoft Health Solutions Group (2008–2012).
  • I was granted an internship at the Washington Hospital Center in 2004 in an unusual way (at that time international internships were rare). Basically, I combined an email web-crawler with a 'for loop' program that I created to get there.
  • In September 2019 and as an IBMer, I joined the Broad Institute of MIT and Harvard as a visiting scientist (press release).
  • I had the honor to give lectures focused on medical informatics and patents as a Distinguished Speaker of the ACM (2018-2021).
  • In 2008, I earned my PhD, focused on human–robot collaboration, from Ben-Gurion University of the Negev in Beersheba, Israel. The path between robotics and informatics may not be logical to most but for me it was straight-forward and inevitable. Following my receipt of an award and a prestigious 6-month internship split between Microsoft Research and the Health Solutions Group, I intended to make a big impact in robotics. However, once I completed my PhD and officially joined the Health Solutions Group at Microsoft (the same group that I interned with) in 2008, I quickly realized that my interest in robotics was waning while my interest in broader medical informatics was growing. As is the case with many others, I could not just leave because I was locked in by immigration restrictions (I was also granted an 'Outstanding Researcher' green card in 2010 that basically tied me to Microsoft), so I had to pivot my career. My new target was the medical informatics team at Microsoft Research, but that never happened given my robotics background. It was at this time that I pivoted to get back on track. More specifically, I went back to academia with a position at Harvard Medical School in bioinformatics, which ultimately led me to my current position at IBM Research.



Top collaborators

Kenney Ng

Kenney Ng

PRS, Accelerated Discovery, Healthcare & Life Sciences | Manager, Center for Computational Health