Bill Marquardt
Bill Marquardt

A laboratory result becomes useful when it reaches the people responsible for care in a form they can understand and use. That makes the connections between healthcare systems part of the diagnostic process itself. As organizations consider new automation and artificial intelligence tools, improving how existing systems acquire, store, share, and interpret information deserves a place at the center of digital transformation.

Recent federal data show both progress and unfinished work. The Office of the National Coordinator for Health Information Technology reported that 76% of U.S. non-federal acute care hospitals participated in all four measured information-sharing activities in 2025. Those activities include sending, receiving, finding, and integrating information. While sending, receiving, and finding improved significantly from 2023, integration remained statistically stable.

These hospital findings provide broader context for the challenge facing diagnostic laboratories. Exchanging a record and incorporating its contents into another system are distinct tasks. The federal measure defines integration as incorporating outside summary-of-care records into an electronic health record without manual entry. That distinction helps explain why a digital connection alone is an incomplete measure of progress.

Shared terminology is another part of the foundation. Regenstrief Institute's LOINC guidance, updated in July 2026, explains that different systems can identify the same test or measurement differently. LOINC provides standardized identifiers that organizations can match to their local codes. However, LOINC coverage is incomplete, particularly for novel and esoteric tests, and laboratories often map codes inconsistently. Some standards therefore already exist, but applying them requires organizations to connect their own terminology to a shared language.

For Bill Marquardt, Founder and CEO of MarquaTech, a healthcare executive leadership and IT consulting firm, this practical work deserves greater attention. Drawing on his experience with healthcare technology, he believes the difficulty of modernization often lies in harmonizing legacy systems, inconsistent and unstandardized data structures, difficulty with inter-system integrations, and the demands of daily operations. Regulatory restrictions compound this problem.

Marquardt explains environments where several systems within an organization use different names or formats for information. Connecting them can require teams to examine individual fields and determine how each should be translated. In his experience, that work can make apparently straightforward technology projects much more involved than leaders initially expect.

"The technology is available," he says. "The challenge lies in integrating it with existing systems while maintaining reliable day-to-day operations."

His concern extends to what happens after a connection is built. Marquardt argues that translating information between systems does not necessarily resolve inconsistencies in the underlying data structures. If organizations continue to store information differently, he believes each new application can inherit another layer of reconciliation work and technical debt. For leaders considering AI, his priority is understanding the information a tool will receive before deciding what responsibilities it should take on.

That perspective also shapes his preference for designing replacement systems around agreed data structures. Marquardt favors a fresh foundation where repeated changes to older systems have become difficult to sustain. However, he acknowledges that laboratories must continue operating throughout a transition. In his view, the challenge is introducing change while preserving the quality and continuity of the services people already depend on.

Responsible AI adoption, he suggests, requires similarly clear boundaries. Marquardt sees potential in tools that help clinicians organize their attention, such as identifying pathology slides that may warrant earlier review. He presents this as a possible supporting role, with a qualified professional remaining responsible for assessing the information and deciding what action to take.

For Marquardt, assigning responsibility also means being clear about what a system is expected to do. He believes healthcare organizations and regulators need a shared understanding of how AI applications should be evaluated for particular tasks. His emphasis is on defining acceptable use and retaining meaningful human oversight as organizations explore new capabilities.

The wider priority, in his view, is coordination across laboratories, healthcare providers, technology developers, and the organizations responsible for regulations and standards. An individual company can improve its own systems, he says, but lasting compatibility depends on choices that others can follow. He believes leaders should treat agreement on data formats and structures as a foundational investment, alongside decisions about software and automation. Without that coordination, he cautions, organizations may find themselves repeating modernization work as their partners move in different directions.

"Shared data standards must be the foundation of healthcare modernization," Marquardt says. "Building that foundation requires a coordinated global effort, with organizations working toward standards they can put into practice together."