How Singapore Is Revolutionizing Public Healthcare Using AI

Traditional AI typically requires specialist knowledge to deploy and is used by enterprises to solve specific business problems, while generative AI entails deploying a large model that is accessible by the general population, observed Andy Ta, the chief data officer at Synapxe.

Ta was speaking at a media panel discussion held at IBM Think 2023 last month. Synapxe is the national HealthTech agency of Singapore, rebranded from Integrated Health Information Systems (IHiS) in July this year.

He shared how Synapxe leverages both AI and generative AI, the various applications of AI within the healthcare sector, and how generative AI has the potential to transform public health services.

Building an AI and data ecosystem

Synapxe is tasked with planning the AI and data analytics ecosystem to support public healthcare, says Ta. “Our focus is on capability development, and we have established a center of excellence for public healthcare, data ethics, and AI.”

“We also work on implementing national platforms and AI projects for public healthcare. Some of our accomplishments include predicting patients at risk of multiple re-admissions in the next 12 months.”

While the team harnesses technology to achieve its objectives, Ta was quick to explain that the focus isn’t around specific large-language models (LLMs) or even the use of AI, but on the outcomes.

“We are agnostic. We don't care about the techniques; we care about the outcome. We are an applied science team. That means we take the science and apply it to healthcare. And that's the most important thing,” he said.

Transforming public health services

What are some AI initiatives that Synapxe has rolled out? Ta shared about efforts to detect diabetic retinopathy, an eye condition that can cause vision loss and blindness in people with diabetes. The team trained an imaging detection model and rolled out the system to all polyclinics so that diabetic patients can have their eyes checked regularly with the support of AI.

The team at Synapxe has also done work around ensuring that patients take their medicine: “For certain medical conditions, caregivers may want to make sure that their parents or loved ones are taking their pills regularly. Now we have this video AI to help them ensure pill adherence.”

One of the strengths of LLMs is its ability to quickly and accurately summarize large amounts of information. However, the general LLMs available today are ill-suited for summarizing healthcare reports. Hence the team designed a specialized model for healthcare practitioners.

“You cannot just summarize a person's health into four, five sentences. We designed something that provides a time range as well as a topic to summarize. So we will summarize the conditions, medication, or problem areas.”

“It can identify gaps in a person's health history, and we will try to highlight key areas so that the doctor won't spend 10 minutes just reading the notes. This gives them the context and most crucial information they need quickly – they can review in more detail if they like.”

AI to help the population

Security is of paramount importance, even more so than financial information which loses relevance over time, says Ta. “We cannot afford our patient’s health data to be leaked. If you are sick, say with diabetes, Next year, you will still have diabetes, as well as the year after. The moment you lose the information, [the privacy is] gone forever.”

For now, the team has plans to leverage LLMs in various ways. For a start, Synapxe is creating a common GPT platform for public healthcare stakeholders to utilize in a safe and secure manner.

“We have created an LLM gateway where we can onboard other models. We are discussing with IBM to include their AI suite from Watsonx to be included on our platform. Stakeholders can explore and try out different models to solve their problems.”

There is also a “guru” model to serve as an expert that healthcare users can interact with, ask questions, and assist in analyzing a case. Ta says Synapxe is also in discussions with IBM to use Watsonx to create an omnichannel hyper-personalized portal to help the Singapore population.

“There are many savvy people using GPT today. But there are even more people who are not savvy, people who don't understand technology, and who are still struggling with using websites. We want to reach out to them. And that’s our mission – to help our population,” he summed up.

Paul Mah is the editor of DSAITrends. A former system administrator, programmer, and IT lecturer, he enjoys writing both code and prose. You can reach him at [email protected].​

Image credit: iStockphoto/Miguel Vidal