Opinion
Why the biggest barrier to scaling AI in South African healthcare isn’t the technology itself, but fragmented, inaccessible and inconsistent data. The article argues that before healthcare leaders ask which AI solution to invest in, they should be asking whether their data environment is ready to support it.
By Henry Adams country director
HEALTHCARE leaders shouldn’t be asking what AI tool to buy first, but whether their data environment is safe and trusted enough to support AI. There is a lot of pressure on healthcare leaders to show how artificial intelligence will improve care, reduce administrative load, and help stretched teams make better decisions.
The expectation is understandable. AI can support faster clinical interpretation, better forecasting, more efficient workflows, and more personalised patient engagement. That promise depends on one condition. AI cannot fix healthcare data it cannot access, understand, or trust.
From my work with healthcare organisations, I have seen that the issue is rarely a lack of data. South Africa’s healthcare system generates enormous volumes of information across hospitals, clinics, laboratories, pharmacies, funders, administrators, and public health programmes. The challenge is that this information does not always move with the patient, the referral, the lab result, or the claim.
In practice, this fragmentation shows up when information does not move cleanly with the patient, the referral, the lab result, or the claim. A receiving facility may lack the patient history needed to act with confidence, while processed laboratory results may not be visible in the workflow where they are needed. Duplicate records, public-private data gaps, and claims fragmentation can weaken the view of care, cost, and outcomes, even in organisations that appear highly digitised.
Getting the data foundation right
Many healthcare AI conversations still start with the tool, the model, or the use case. Those questions are important, but they only become useful once leaders understand whether the data environment can support AI safely, consistently, and at scale.
Trusted data is not simply data stored electronically. It is accurate, governed, secure, traceable, and available in context. It can move between systems without losing meaning, support auditability and accountability, and remain usable for clinicians, administrators, and decision-makers without creating another layer of complexity.
A weak data foundation creates inefficiency, but its impact can go further. It can affect clinical decisions, delay services, weaken governance, and erode public confidence. In healthcare, trust is an operating requirement.
Building for connected care
South Africa’s healthcare environment adds another layer of complexity. Public and private providers operate under different pressures, yet both face the same challenge: connecting information across systems, facilities, and workflows to improve patient care, operational efficiency, and organisational resilience.
Whether the goal is improving referral pathways, reducing turnaround times, supporting AI-enabled services, or preparing for future healthcare reforms such as National Health Insurance (NHI), success depends on the same foundation. Data must be able to follow the patient, the process, and the decision.
This does not mean every organisation needs to replace every system. In most healthcare environments, that would be unrealistic, expensive, and disruptive. A more practical approach is to create a trusted data layer that connects existing systems, standardises data exchange, strengthens governance, and makes reliable information available wherever it can create the greatest value.
Creating operational visibility
One South African example is forensic chemistry, where digital laboratory infrastructure must support blood alcohol and toxicology workflows, strict auditability, secure chain of custody, and reporting that provides teams with visibility into turnaround times and backlogs. The broader healthcare lesson is that trusted data becomes most valuable when it improves operational confidence.
In this context, digitised processes make a difference. The larger requirement is an information foundation that allows the organisation to see itself clearly. When data is connected and governed, leaders can identify bottlenecks, monitor performance, strengthen reporting, reduce manual work, and make better operational decisions. Once that foundation is in place, AI becomes more useful because it is working from information with context and integrity.
Without that foundation, AI risks becoming another isolated layer on top of a fragmented environment. It may work in a pilot, where the data is carefully selected and the use case is narrow. Those demonstrations may be impressive, but scaling AI across healthcare requires interoperability, security, data quality, governance, and operational reliability.
For South African healthcare leaders, I believe the immediate priority is to assess readiness and determine whether the organisation has the data environment needed to adopt AI responsibly. That means knowing what data is available, where it comes from, how it has changed, whether it can be shared securely, and whether it follows recognised healthcare standards. It also means being able to govern that data in line with privacy, compliance, and accountability, while ensuring it can support clinical and operational decisions in real time.
Algorithms will only improve healthcare if the information beneath them is reliable enough to support clinical and operational decisions. In South Africa, that makes the data foundation urgent. The country cannot afford digital ambition built on disconnected information, especially when the goals are better outcomes, stronger systems, and more responsive care. AI has a role to play, but its value will depend on whether healthcare organisations can trust the data it uses.
***Henry Adams is the country director at at InterSystems, South Africa.
F.O
Tags: Hwalthcare AI South Africa