.png)
Bridging AI with Healthcare Needs 🔗🫱🏻🫲🏻
There was a point in my healthcare data science journey where I remember thinking, “Am I still a data scientist, or am I also becoming an application developer, workflow designer, product thinker, and translator?”
It was slightly uncomfortable, but also exciting.
I started out closer to the world of applied research, analytics, and modelling. My instinct was to study the problem carefully, understand the data, and make sure the analysis or model was rigorous. But as I became more involved in digital transformation work, I realised that building useful AI in healthcare requires much more than a technically sound model.
Healthcare data is complex. AI models must be developed carefully, validated safely, and used appropriately. But the work cannot stop at looking at data, building models, or creating applications. The most important question is still: who is this for, and will it actually help them?
Today, my work sits across data science, AI, and application development. As a technical lead supporting both the Digital Think Tank DA SWAT team and the AI CORE department, I help translate ideas into practical tools that support the hospital.
I see my role as sitting between technical possibility and healthcare reality.
On one side, there is the data science and AI work: understanding messy healthcare data, developing models, validating outputs, thinking through accuracy, safety, and governance, and ensuring solutions are clinically appropriate.
On the other side is the implementation work: understanding what frontline users actually need, how they work, what constraints they face, and how a tool can fit naturally into their day without adding burden.
That has been one of my biggest lessons. A model can perform well on paper, and an application can work technically, but if it does not fit the workflow or feel useful to the people using it, it will not create meaningful impact. In healthcare, technology must serve the workflow, not the other way around.
The spark came from realising that many problems in public healthcare were not due to a lack of effort from staff. Very often, people were already working extremely hard. The challenge was that they were navigating fragmented systems, repeated manual steps, and information that was not always easy to access at the point of need.
As a data scientist, my comfort zone was to analyse the problem and produce insights. Over time, I realised that insights alone were not enough. If we could see the problem clearly, perhaps we also had a responsibility to help build solutions that address it.
A large part of my journey has been learning to operate across both worlds: the rigorous data science domain and the practical application-building space. Some ideas sound good in discussion, but once we examine the workflow more closely, they need to be reshaped. Some prototypes work technically, but do not yet feel natural for users. Sometimes the feedback is very direct — and that is exactly what we need.
That feedback forces us to ask whether we are truly solving a frontline problem, or just building something that looks impressive from a technical point of view.
I have also learned that building solutions in public healthcare is very much a team sport. Clinicians help us understand what matters clinically. Nurses and operational users show us what will or will not work on the ground. Governance and leadership create the conditions for safe experimentation. Technical teams turn ideas into something real.
For me, meaningful impact is seeing AI move from an abstract concept into something users can react to, challenge, test, and shape. That process matters because it turns digital transformation from something done to users into something built with them.
The goal is not to replace human expertise. It is to support it. Healthcare staff already carry a heavy cognitive and emotional load. Good digital tools should reduce friction, not add to it. They should help people spend more time on the parts of care that require judgement, empathy, and experience.
This journey has taught me that courage in digital transformation is not always dramatic.
Sometimes, it is simply the courage to try something before you feel fully ready. It is the courage to ask basic questions, receive difficult feedback, admit that the first version is not good enough, and keep improving it with the people who will actually use it.
My message to colleagues would be: you do not need to be a “digital person” to contribute to digital transformation. If you understand a problem deeply, your perspective is already valuable. Some of the best ideas come from people closest to the work.
For me, courage is not about being fearless. It is about being willing to step outside your comfort zone, learn in public, listen to different perspectives, and build something useful together.
Written by Jonathan Yeo, Lead, AI Engineer, AI CORE, NTFGH