Back

Robotic nursing assistants, innovative AI initiatives like DAISI unveiled at Imagine AI conference

The future of healthcare shimmers with promise at the National University Hospital, where a robotic nursing assistant will be introduced; while Duke-NUS launched the Duke-NUS AI + Medical Sciences Initiative, or DAISI, to develop novel AI solutions to real-world clinical problems. These were two of new several innovations unveiled during the sixth Imagine AI conference, which ran from 5 to 6 December at the Sands Exhibition and Convention Centre.

DAISI brings together clinicians and AI and data science experts to innovate in the healthcare AI space, focusing on areas including cellular data science, immunology, quantum computing and AI governance and ethics.   

Sparking off the discussions for the two days, Guest-of-Honour Infocomm Media Development Authority Chief Executive Mr Lew Chuen Hong challenged the audience to look beyond the hype of AI and focus on understanding the problem so that the right approach can be used to address it. Because AI is like a toolbox, it contains many different tools, each suited to a specific purpose.

“If I may borrow this term, just as you use drugs in cocktails, I think you should use AI wisely as a cocktail," he said.

To be able to use such a toolbox to improve healthcare, Mr Lew added, doesn’t require training more computer scientists, but rather requires nurturing a bilingual generation of doctors, who are equally well versed in the understanding of human biology and physiology and technology.

The need to train bilingual doctors was also a cornerstone of the presidential panel discussion, during which leaders from the six main co-organising institutions shared their insights with the more than 600-strong audience of AI, healthcare and industry experts from around the world.  

Duke-NUS Dean Professor Thomas Coffman who was on the panel, echoed this highlighting that medical schools need to ensure that students understand AI, how it works, what it can do as well as what the limitations and ethical considerations are.

At the same time, AI can play a huge role in medical education, in areas like simulations, as well as in research where it is revolutionising data analysis and drug discovery. But looking ahead to 2030, Prof Coffman predicted that one of AI’s biggest impacts will be as a companion to medical students throughout their journeys in medicine:

“Our medical students will have a virtual concierge that will help them through the admissions process, their learning, assessment, logging of procedures, as well as help them negotiate the process of getting into residency,”

This concept of agentic AI had also been the focal point of the opening panel discussion, titled “Your AI-enabled doctor will see you now!” with healthcare innovator and DeepLearning.AI founder Dr Andrew Ng beaming in from the US to join fellow panellist Professor Ngiam Kee Yuan from the National University Health System and moderator Mr Steven Chia from Channel NewsAsia on stage.

The two panellists agreed that agentic AI is the most important technology trend in AI. In this use of AI, developers build a workflow into a generative AI model that returns a first response into the system to critique and improve until the result has reached the desired level.

“Agentic AI technology is exciting because developers are starting to use AI in a much more iterative way,” said Dr Ng, who added that using an agentic AI workflow may be a bit slower but produces much better outputs. It also opens the avenue for different AI agents to interact with each other.

As academics, healthcare professionals and industry experts explore how to automate tasks or incorporate AI into clinical workflows, collaboration is key, noted Associate Professor Liu Nan, director of DAISI, who opened the second day.

“We cannot afford to work in silos,” he emphasised, adding that AI is more than a technology, it is a whole field that needs to be filled with domain knowledge so that the solutions that are developed are effective, safe and ethical, a topic that was expanded on during the closing panel of the conference on “Ethical considerations in AI for medicine”.

The launch of large language models into the public domain may have caused much excitement and stirred great hype, but it was clear from the discussions that AI will continue in more measured steps to shape healthcare as it embraces AI.

One day in the not-too-distant future, AI will regularly manage tasks such as notetaking, first-line screening for common conditions as well as providing advanced triaging support and astute differential diagnoses; leading to a partnership much like the one between Ironman and his digital assistant Jarvis, where the human’s decision-making power is augmented by AI and our time freed up to upskill and focus on the interaction that matter. It just won’t happen overnight.