PARTNER

From data to decisions: Why healthcare AI readiness depends on more than technology

Healthcare institutions are under growing pressure to turn artificial intelligence (AI) ambition into practical value. But in complex care environments, AI adoption cannot be treated as a standalone technology shift.

At HealthTechX Asia, leaders from National University Health System, Thoughtworks and the private healthcare sector came together for a panel discussion on “From Data to Decisions: Turning Healthcare Data into Strategic Advantage”. Image: Canva

The conversation explored what it takes for healthcare organisations to move from AI experimentation to trusted adoption.

One theme stood out: AI readiness in healthcare starts long before the first use case goes live.

It depends on whether institutions have the foundations to capture the right data, govern it responsibly, embed it into workflows and act on insights with confidence.

Modernisation has to start with outcomes

For many organisations, modernisation is still approached as a technology project. Legacy systems are replaced, platforms are upgraded and infrastructure is moved to the cloud.

But in healthcare, the more useful starting point is the outcome the institution is trying to achieve.

AI initiatives often struggle when data is fragmented across systems, workflows and business units.

Without a clear view of what data is being captured, where it comes from and how it supports decision-making, even well-funded programmes can struggle to move beyond experimentation.

This is why modernisation needs to be anchored in business and care outcomes from the start.

Institutions need to ask what problem they are trying to solve, what decisions need to improve and what data is required to support those decisions.

Healthcare organisations also do not need to wait for every modernisation effort to be complete before experimenting. Targeted use cases can run in parallel with infrastructure improvements, helping teams demonstrate value, build confidence and create momentum for broader change.

Governance should enable adoption, not slow it down

As AI becomes more embedded in healthcare workflows, governance cannot be treated as a final checkpoint before deployment.

It needs to shape how systems are designed, tested, used and monitored from the beginning.

Clear governance can help teams move faster because it defines where they can act, what boundaries they must observe and how risks should be managed.

This becomes even more important as AI systems become more agentic. A research agent, a scheduling agent and an agent involved in prescription-related workflows do not carry the same level of risk.

Each requires a different level of control, depending on what the agent is allowed to do and whether its actions can be reversed.

The shift is from governance as policy to governance as operating discipline. Healthcare institutions need to anticipate what can go wrong, design mitigations early and ensure there is accountability for the actions AI systems support.

Data trust begins at the point of care

Good AI depends on good data, but healthcare data quality is not only a technical problem.

It is shaped by how data is captured, how workflows are designed, how consent is managed and who takes ownership of the data once it enters the system.

Patient consent, for example, cannot be treated as an afterthought. It needs to be considered before data is captured and built into the patient journey from the beginning.

Once data is captured, lineage becomes critical: teams need to know where the data came from, how it has been transformed and whether it can be trusted for AI use.

Ownership also matters. Data governance cannot sit with technology teams alone. The clinical or business domain that understands the data must play a role in deciding how it is classified, used and exposed.

As AI produces more analysis, content and recommendations at speed, healthcare organisations will also need to strengthen the human capability to validate what AI generates.

People need to be able to question outputs, understand limitations and take responsibility for the decisions that follow.

From AI adoption to AI readiness

Healthcare AI adoption will not be defined by the number of tools an institution deploys. It will be defined by whether those tools can be trusted in the environments where they are used.

That trust depends on more than technical performance. It depends on whether data is reliable, workflows are ready, governance is embedded and people remain accountable for the decisions AI supports.

The next phase of healthcare AI will belong to institutions that can connect these foundations: modernising with purpose, experimenting responsibly and building the governance and human judgement needed to turn data into decisions.

-------------------------------------------

About Thoughtworks:
At Thoughtworks, we help healthcare organisations and regulators build the platforms, practices and governance foundations needed to move from AI ambition to trusted, accountable adoption.