How Singapore hospitals are pioneering AI

By Nurfilzah Rohaidi

Solutions do not have to be ‘hugely expensive’ to make an impact on patients’ health, according to the Smart Health Leadership Centre in Singapore.

It’s a cool, breezy evening, and you are out for a run. The fitness tracker strapped to your wrist beeps. It has detected a change in your gait - forecasting a possible injury in a few weeks.

AI-enabled devices like these are already possible today, said Tamsin Greulich-Smith, Chief of the Smart Health Leadership Centre. She works closely with hospitals in Singapore to help them use AI, data analytics, design thinking techniques and what she calls ‘human-centred design’.

At the recent Data Governance and AI Summit presented by GovInsider and SAS, Greulich-Smith shared how hospitals can approach AI not as a huge monster to tame, but as an enabler and a means to boost patient outcomes. “One of the biggest questions I get asked is how do I know what to invest my money in?” she remarked. "We always go back to what impact you're trying to create.”

‘Start small, think big’

The centre has been using AI to innovate across Singapore's hospital network, targeting problems such as fall prevention, vitals monitoring, stroke rehabilitation and social isolation. But when it comes to AI, hospitals have concerns around cost, scale, complexity, integrating with legacy systems - and of course, the ethics of it all, she noted.

Yet, done right, it is possible to achieve meaningful impact on people’s health and wellbeing despite these barriers.

One recent project involving AI voice-enabled assistants only cost S$6,000. “We were trying to explore the role of AI in trying to nudge behaviors in socially isolated seniors,” Greulich-Smith explained.

Yishun Community Hospital in Singapore had seen how social isolation in recently-discharged elderly often led to recurring visits to the hospital. “Social isolation is linked to lower health outcomes across the board,” she continued.

Her team worked with social services agency Lions Befrienders to introduce an off-the-shelf AI device into seniors’ homes. In this case, they chose Amazon’s Alexa. They explored using the devices to remind these seniors to eat, take their medication, or nudge them into joining social activities.

This small trial had surprising results: “We noted that the seniors unexpectedly built a little relationship with Alexa,” Greulich-Smith remarked, noting that these patients would say goodbye to the voice assistant before leaving their homes.

Her team also made sure that the data collected by these devices could be shared easily between the healthcare provider and the social services agency, she added. “We were looking to stitch data flow up around the individual patient.”

Gamification for rehab

The centre also sought to tackle stroke rehabilitation using AI. This project involved “health by stealth”, where the centre explored how to use gamification to make repetitive rehab exercise for stroke patients more fun.

The platform uses an infrared-based camera to track the joint movements of patients’ hands and fingers, and displays video games to allow patients or stroke survivors to play and feel engaged, said co-presenter Dr Matthew Chua, who is a lecturer and consultant for medical and cybernetics systems at the centre.

This way, it was possible to track the incremental improvements of stroke patients’ progress and recovery. “We needed to actually track and quantify how much motion or angle of grasp the patient has recovered through this few weeks of exercises,” Chua explained. “Everything looks the same to the naked eye.”

The system is now ready for public use at a local shopping mall, and was made possible by government funding of half a million dollars - “relatively cheap” in the grand scale of things, Greulich-Smith pointed out.

Often, hospitals worry that they cannot afford AI solutions, she said. They believe that “healthcare AI means hugely expensive solutions that I don't know if I can afford to prioritise when I have so many growing demands on my health care services, and I don't even think I have the skills”. But these are just two examples of projects that created meaningful impact without blowing budgets, she noted.

Three tips for hospital leaders

She went on to share three pieces of advice for hospital leaders seeking to explore AI to change outcomes. First, “collaboration is your friend in the digital world”, Greulich-Smith asserted. The first step is to find the right partners to share the journey with you.

Next, “think about data”, Greulich-Smith said. Hospitals these days are “swimming” in data, oftentimes existing only in siloes. She believes that hospitals need to start thinking about what data is relevant to the end goal, and then “start cleaning that up and working out where it's stored and how it's collected so you can use it”.

And finally, she hopes that hospitals can also be “trust-aware” when designing solutions that use AI. “It can build trust and decision making, it can help us achieve things that we couldn't do on our own. However, there is a risk that we've placed all that trust in it, and that we still need a human eye in there,” Greulich-Smith mused.

The benefits can change how we approach and think of healthcare. Family background and environment are key factors, and by making predictions through monitoring devices and data analytics, for instance, physicians can understand future health risks better and “that allows us to be more preventative”. “It allows us to be more personal than ever about health,” she noted.

The potential for AI to transform healthcare is already right in front of our noses. Now, the next frontier to overcome, as Greulich-Smith put it: “We want to try to break down some of the myths that AI needs to be something really big and terrifying”.