More than 700 doctors, data scientists, data engineers, software engineers and innovators gathered at the Healthcare AI Datathon and Expo 2019 to explore how artificial intelligence (AI) can be used to tackle healthcare challenges.
Jointly organised by the School, National University Health System (NUHS) and the Massachusetts Institute of Technology, the event held on 17-21 July comprised conferences, seminars, workshops and exhibitions featuring prototypes using machine learning algorithms to improve patient outcomes.
Event highlights video from NUHS
The Datathon held on 19-21 July brought together clinicians, data scientists and statisticians to work together to come up with solutions in using AI for critical care, general care and medical imaging.
“The whole objective of this datathon is to bring together physicians and to join forces with the data scientists,” said Assistant Professor Feng Mengling, who led the Datathon.
He added, “We noticed from our previous events that a lot of these collaborations are good, but a lot of this research ends up in publications and stays at the academia environment. However, this year we want to do something better, something more. So what we did differently from the previous year is that even after the datathon, we will continue to offer mentorships and offer funding opportunities to encourage the teams to continue to work together and bring their research projects to not just end up in publications but rather some real products or solutions that can impact our healthcare system.”
Dr Feng collaborates closely with Dr Ngiam Kee Yuan, Group Chief Technology Officer, NUHS to develop AI solutions to augment the way clinicians work and allow patients to receive better care.
“We are talking about a complete rethink of how we approach healthcare,” said Dr Ngiam. “Take for example waiting times, instead of waiting longer, what if you didn’t need to come in in the first place? You could have Tele-Health consultations at home.”
Dr Feng also shared about FathomMammo, an AI tool that can read digital mammograms and diagnose breast cancer. The tool can assist radiologists by performing the first set of readings and detect cancerous tissues from the mammograms, draft a preliminary report on the findings, and suggest diagnoses.
“We found that over 70 per cent of the time, the radiologist agrees with the recommended diagnosis by AI. That can potentially cut down the 20 to 30 minutes of reading time to just a couple of minutes,” said Dr Feng.
Media coverage: