The Diagnosis

The Technology Is Not the Problem. The Fit Is.

Most digital health projects do not fail because the technology is bad. They fail because the technology was built for a different environment and nobody said so until the go-live.

I want to talk about a distinction that I think explains more health system implementation failures than any other factor. It is the difference between technology that is bolted on to clinical work and technology that is built in to it.

Bolted on means a system is implemented alongside existing workflows. Clinicians are trained on it, expected to adapt to it, and assessed on their adoption of it. The technology is complete. The integration with how care is actually delivered is partial.

Built in means something different. The technology was designed around how clinicians work: the decisions they make, the sequence in which they make them, the constraints they are operating under. The clinician does not adapt to the system. The system adapts to the clinician.

That distinction matters because it predicts outcomes. Research published this year found that clinicians using a workflow-aligned AI decision support tool made 16% fewer diagnostic errors and 13% fewer treatment errors compared to those without access. The same research was clear about why: those results only emerged when the tool was designed around clinical workflow. When similar tools were deployed without that alignment, bolted on rather than built in, adoption plateaued and the clinical benefits did not appear.

The Caribbean Context

For health systems in this region, this matters more than anywhere else.

Our clinical environments are not high-resource, fully staffed, consistently connected settings. They are under-resourced, intermittently connected, and staffed by clinicians managing patient volumes that leave very little margin for technology friction.

Digital health tools designed for environments with reliable infrastructure, deep IT support, and dedicated implementation teams do not translate cleanly into that reality. The failure pattern is predictable. Systems that work in controlled conditions break under real operating pressure. Clinicians build workarounds. The workarounds become the standard. The technology sits unused.

What works in these environments is technology that was designed with the constraint in mind, not retrofitted to it afterward. Tools that function offline when connectivity drops. Interfaces that take minimal training to use. Failure protocols embedded in the implementation from day one, not added as an afterthought.

A Governance Question

Here is what I want health system leaders to take away from this.

The built-in versus bolted-on question is not primarily a technology procurement issue. It is a governance issue. The board or executive team approving a digital health investment should be asking: has the implementation plan accounted for what this environment actually looks like? Has clinical input shaped the design, not just the training? Is there a plan for what happens to patient care when the system fails?

Most procurement processes evaluate what a system can do. Far fewer ask what happens when it cannot.

A $300 billion global market is generating digital health technology at extraordinary scale. The Caribbean's task is not to keep pace with that scale. It is to be precise enough about its own clinical environment to select and implement the fraction of that technology that was actually designed for conditions like ours.

Build it in. Or leave it on the shelf.