How healthcare firms can fix data governance problems

Healthcare organizations have spent the past two years racing to evaluate and deploy artificial intelligence. From revenue cycle management and patient engagement to clinical documentation and analytics, health systems increasingly view AI as a tool that can help relieve workforce pressures while improving operational performance.

Yet as enthusiasm grows, a more fundamental challenge is moving into the spotlight: the quality of the data that powers those systems.

A recent S&P Global report suggests many healthcare organizations remain concerned about the integrity of the information flowing through their systems. For Joe Hickey, vice president of provider markets at Verato, the issue is not simply a technology problem. It is a business, operational and patient experience challenge that becomes more consequential as AI adoption expands.

The concern stems from a basic reality. AI systems are only as effective as the data they receive. When patient records contain inaccuracies, duplicates, mismatched identities or incomplete information, those flaws can influence everything from consumer engagement to automated decision-making.

Data friction patients can feel
Consumers increasingly expect healthcare organizations to deliver the same seamless digital experiences they encounter in banking, retail and travel. According to Hickey, healthcare often falls short when identity and demographic data are inconsistent across systems.

The consequences can be highly visible to patients. Repeated requests for the same information, registration errors and communication breakdowns create frustration that can damage trust in an organization.

According to findings cited by Hickey, nearly 85% of consumers say they would consider switching providers after repeated identity or data errors. Meanwhile, 81% report having to provide the same personal or health information multiple times across visits, departments and systems.

Healthcare executives recognize the challenge as well. Hickey pointed to research showing that 81% of providers and payers agree they cannot deliver personalized care or communications without complete and accurate consumer data.

“AI, as powerful as it is, cannot function to the height of its abilities with contradictory, missing or inaccurate data,” Hickey said. “This creates problems with access, care and revenue.”

When AI inherits bad information
The healthcare industry’s AI conversation often focuses on model performance, governance frameworks and regulatory oversight. Hickey argues that organizations should devote equal attention to the underlying information feeding those systems.

“AI, for all its amazing capabilities, is still subject to the most fundamental rule of data processing: garbage in, garbage out,” he said.

The challenge becomes particularly significant when flawed data is used to train or inform AI systems. Inaccurate information can produce mistakes that are not immediately obvious, creating downstream effects across multiple workflows and applications.

“AI that has been trained on faulty datasets will continue to make errors with the result that a single data problem can reverberate through multiple applications undetected, compounding problems for users,” Hickey said.

Those situations can create a disconnect between perceived and actual performance. Users may blame an AI application when the underlying issue is poor-quality data. As organizations scale AI initiatives, Hickey said, addressing data integrity before implementation becomes increasingly important.

The financial cost of mismatches
The implications extend well beyond technology performance.

The S&P Global report found that 84% of healthcare organizations believe data mismatches already contribute to lost revenue. While manual remediation can be expensive, Hickey said the broader financial impact touches multiple parts of the enterprise.

Patient access is one example. Scheduling difficulties, registration issues and communication failures can discourage patients from engaging with providers or following through on care plans. Lost engagement can ultimately translate into lost revenue.

Within clinical operations, mismatched records can complicate care coordination and follow-up efforts. Administrative departments face additional challenges when demographic, identity or insurance information is inaccurate.

“Mismatches are particularly costly in revenue cycles and administration, where 83% of providers and payers agree that data quality issues hinder the effectiveness of their marketing and CRM processes,” Hickey said.

Billing errors, duplicate claims, coverage discrepancies and reimbursement delays can all emerge from inaccurate identity data, adding friction to already complex revenue cycle processes.

Building the foundation before scaling AI
Healthcare organizations face significant pressure to demonstrate value from AI investments. At the same time, many leaders believe the technology can help address staffing shortages, financial constraints and growing consumer expectations.

That urgency, Hickey said, can sometimes lead organizations to move faster than their data infrastructure allows.

“Healthcare’s rush to embrace AI without first securing data readiness is like people tossing aside the instruction manual in favor of hands-on learning,” he said.

The challenge is particularly acute because AI initiatives often begin as pilot projects before expanding across departments and workflows. If foundational data issues remain unresolved, organizations may struggle to achieve expected returns or maintain trust in the technology.

A reliable identity foundation
For Hickey, the solution begins with establishing a reliable identity foundation that can accurately connect information across disparate systems and ensure that records correspond to the correct individual.

“Closing this gap requires more than just incremental improvements,” he said. “It calls for a unified, trusted identity foundation that links data across systems and ensures that every interaction connects to the correct individual.”

As AI becomes more deeply embedded in healthcare operations, the industry’s long-standing data quality challenges are taking on new urgency. Advanced algorithms may capture the headlines, but their success still depends on something far less glamorous: accurate, trustworthy information.

Even then, Hickey said, oversight remains essential.

“And, of course, AI needs to be continuously monitored, not only by keeping the proverbial human-in-the-loop, but also through automatic measures and safeguards,” he said. Healthcare IT News