How historical data is reframing the AI conversation in healthcare | Viewpoint

Artificial intelligence is dominating the healthcare conversation, but not always in the most helpful ways.

At conferences, in vendor pitches, and across leadership roundtables, the term is applied broadly to everything from automation tools to large language models. As the noise increases, the real question for healthcare leaders is no longer if AI should be adopted, but how it can be used to solve real problems and what foundational steps must be taken to support it.

One of the most important, and often overlooked, foundations is an organization’s historical data. For all the attention paid to emerging tools and platforms, the real engine behind effective AI isn’t new code, it’s historical context. And in healthcare, that context lives in legacy systems.

Building the future on the past

AI isn’t a one-size-fits-all solution. It’s a family of technologies – machine learning, predictive analytics, natural language processing – that can be applied in very different ways depending on the problem being solved.

But none of these tools can generate meaningful results without data that is accurate, structured, and sufficiently rich in volume and history.

This is where historical data plays a central role. Systems retired years ago often contain decades of patient histories, clinical transactions, and financial records – an invaluable source of intelligence if accessed and prepared properly.

Unlocking this data transforms it from a compliance requirement into a strategic asset, ready to fuel innovation.

Strategy before technology

Successful AI implementation starts with use cases, not technology. Whether the goal is to support overburdened revenue cycle teams, enhance population health analytics, or improve clinical decision-making, the application must drive the tool, not the other way around.

And those use cases are only as strong as the data beneath them. For example, training a model to predict readmission risk is far more effective with years of patient discharge data. Identifying patterns in patient behavior or payment likelihood works best when historical financial data is accessible and linked across systems. In this context, the value of active archiving – bringing structured, historical data into reach – becomes a key enabler of AI strategy.

Responsible AI starts with responsible data

With growing capabilities comes growing responsibility. AI must be adopted thoughtfully, with governance models that protect patient privacy and institutional trust. But responsibility also means ensuring that the data being fed into AI systems is clean, complete, and free from bias—criteria that can only be met when historical data is intentionally integrated into the broader data strategy.

It's also important to ask the hard questions: Where does the AI engine live? How is the data being stored and secured? Are there clear guardrails in place to prevent misuse or unintended consequences?

Ultimately, responsible AI adoption is not just about compliance. It’s about creating systems that serve people – augmenting staff, not replacing them – and delivering real value to clinical, operational, and financial teams.

Cutting through the noise

Healthcare leaders today face an overwhelming number of AI conversations. The pressure to explore, invest, and innovate is constant but the risk of jumping too fast, or without a strategy, is high. Many are being asked to make decisions without full clarity on ROI, timeline, or impact.

That’s why internal alignment is just as critical as external due diligence. AI tools being considered for patient financial services should involve those end users in the evaluation. If the tool is meant to improve productivity, it must be tested against real workflows—not theoretical ones.

And when assessing value, healthcare organizations should move beyond promises of cost-cutting. In most cases, the goal isn’t to reduce staff – it’s to help existing teams accomplish more with the resources they already have. AI can deliver measurable gains in speed, accuracy, and insight—but only when those metrics are clearly defined and tied to specific use cases.

From archiving to innovation

Perhaps the most powerful shift in AI strategy today is the reframing of historical data. What was once treated as a liability – old software, inactive systems, maintenance costs – is now being recognized as a competitive advantage. Active archiving has evolved from a back-office necessity into a strategic function that supports AI readiness.

Through application rationalization, data extraction, and integration across platforms, organizations are creating cleaner, more connected datasets. These datasets, in turn, provide the foundation for predictive models, intelligent automation, and future-forward technologies.

The breadth and depth of historical healthcare data represent more than a record of the past—they offer a roadmap for the future. And for AI to deliver on its promise, that past must be accessible, structured, and put to work.

Jim Jacobs is president and chief executive officer of MediQuant.