How can project management determine whether AI can expand healthcare access

5 days ago

5 min read

Updated: 1 day ago

How can project management determine whether AI can expand healthcare access

And what Horizon 1000 tells us about managing AI-enabled transformation

Bill Gates' recent article, Expanding access to health care through AI, describes Horizon 1000, a partnership involving the Government of Rwanda, the Gates Foundation and OpenAI. Its goal is to support the use of AI across 1,000 primary healthcare clinics and their surrounding communities by 2028. The ambition is compelling: help overstretched health workers provide better care to more people, while keeping clinicians at the centre of decision-making.

It is also a useful case study in why the next phase of AI will be won or lost through delivery. The technology may be powerful, but a model does not change a health system by itself. Benefits emerge only when technology, clinical practice, workforce capability, governance, infrastructure and community trust move together.

That is where project, programme and portfolio management matter.

The article is about AI, but the challenge is transformation

The healthcare problem described by Gates is not simply a shortage of information. It is a shortage of capacity within a complex system. Health workers face overwhelming demand, limited administrative support, uneven infrastructure and difficulty accessing current clinical guidance. AI could help with triage, documentation, decision support and access to reliable knowledge. Yet introducing it changes how work is performed, how decisions are made and where accountability sits.

A narrow technology project might measure whether a tool was configured, tested and deployed. A transformation program must ask harder questions: Did care become safer? Did workers gain useful capacity? Did patients obtain quicker access? Were disadvantaged groups included? Did the solution strengthen the health system, or create a dependency that cannot be sustained?

This distinction matters in every sector. AI projects are often presented as software implementations, when they are really operating-model changes with technological components.

How AI changes project management

AI will alter both the work project professionals perform and the environment in which projects are delivered.

First, routine project work will become faster. AI can help analyse information, prepare drafts, identify patterns, test scenarios and maintain project controls. This should free project professionals to spend more time on judgement, relationships, decisions and intervention. The value will not come from producing the same documents more quickly. It will come from improving the quality and speed of decisions.

Second, plans will need to accommodate faster learning. AI performance is probabilistic, data-dependent and sensitive to context. A tool that performs well in one language, clinic or workflow may not perform equally well in another. Delivery therefore needs controlled experimentation, evidence-based stage gates and rapid feedback from frontline users, rather than a long design phase followed by a large-scale release.

Third, project success measures must expand. Time, cost and scope remain important, but they are insufficient for AI-enabled change. Measures should include clinical or service outcomes, safety, equity, adoption, workforce impact, trust, data quality, accessibility and the continuing ability of humans to question or override the system.

Finally, assurance must become continuous. Model behaviour, data, cyber threats, regulation and user practices can change after implementation. Approval at go-live is not the end of governance. Monitoring, escalation, periodic validation and clear accountability must continue throughout operation.

How project management can help AI deliver real healthcare benefits

1. Begin with the outcome, not the tool

A strong business case should define the healthcare problem, the people affected and the measurable outcomes sought before selecting a solution. For Horizon 1000, success is not the presence of AI in 1,000 clinics. It is improved access, quality and workforce capacity across those clinics and communities. Clear outcome ownership also prevents the program from becoming a collection of disconnected pilots.

2. Design with health workers and communities

The article emphasises that AI should support health workers, not replace them. That principle needs to be translated into delivery practice through co-design, workflow observation, user testing and feedback. Clinicians, administrators, patients and community representatives should shape requirements and trade-offs. This improves usability and exposes risks that a central technical team may not see.

3. Govern according to consequence, not only cost

A relatively inexpensive AI tool can influence high-consequence decisions. Governance and assurance should therefore reflect the potential impact on patients, workers, privacy and public trust. Decision rights must be explicit: what can the AI recommend, what must a clinician decide, when is human review mandatory, and who acts when performance falls outside tolerance?

4. Deliver in stages and learn deliberately

A staged approach can move from discovery and controlled trials to broader implementation only when agreed evidence thresholds are met. Each stage should test technical performance, workflow fit, user adoption, safety and equity. Lessons should be captured across sites so the programme learns as a system, rather than asking every clinic to solve the same problem independently.

5. Treat readiness as a deliverable

Deployment should depend on more than technical completion. Each site needs suitable infrastructure, data, training, support, local leadership and time to absorb the change. Portfolio management can sequence releases around operational capacity and other initiatives, reducing the risk that several well-intentioned changes overwhelm the same workforce.

6. Manage benefits beyond go-live

Benefits often emerge after a project team has moved on. Named business and clinical owners should remain accountable for adoption and outcomes, supported by a baseline, benefit measures and post-implementation reviews. Where results differ across clinics or population groups, the program should investigate why and adapt.

7. Build sustainability from the start

Long-term value depends on funding, local capability, vendor arrangements, support, model maintenance, data stewardship and knowledge transfer. A successful pilot that cannot be maintained or scaled is not a successful transformation. Project management provides the discipline to design transition and sustainment before external support reduces.

A new role for the project professional

The implication is not that project managers need to become data scientists or clinicians. They do need enough AI literacy to ask informed questions, recognise uncertainty and bring the right expertise into decisions. More importantly, they must remain integrators of the whole system.

In AI-enabled transformation, the project professional connects strategy to delivery, technology to people, experimentation to governance, and outputs to sustainable outcomes. This role becomes more important as technology accelerates, because speed without integration can scale harm as readily as benefit.

The real opportunity

Horizon 1000 is a powerful reminder that AI's value is not inherent in the technology. Value is created through thoughtful implementation in a real human system. Project management supplies the structure for that journey: a clear purpose, inclusive design, staged learning, responsible governance, readiness, benefits realisation and sustained capability.

If AI is to make healthcare more accessible, project management must do more than deliver the technology. It must help ensure the right change reaches the right people, works in practice and continues to create value after the project is complete.

Next steps

The PMLogic® team extensive experience in healthcare and other transformation programs. As a for purpose strategy implementation educator and specialist this experience is brought to our projects through experienced practitioners, so if your current approach could be improved, and most can, please contact one of the PMLogic team to discuss how we can help you and your organisation improve the value it creates.