AI in Healthcare Hinges⁤ on Data Quality and Transparency, Experts Say

Updated June 22, 2025

LAS VEGAS – Healthcare IT leaders at the ViVe 2025 Conference addressed the critical role of data quality and security in the age of artificial intelligence. During the session⁤ “Data⁣ S-A-F-E-T-Y, Find Out What it Means⁣ to AI,” panelists emphasized that AI’s potential to improve healthcare delivery depends on maintaining patient trust and data integrity.

Leah Miller, Chief Clinical Applications & Data Officer at CommonSpirit⁣ Health, stressed⁣ the importance ‍of a solid data foundation. “Everything in ⁣our future as an industry is centered on data,” Miller said. “But the challenge is, how do we ‍get data to move, how do we escape archaic infrastructures, and how do we create a common language of data?”

Miller noted that despite decades of effort⁢ to modernize electronic medical records (EMRs), many healthcare organizations still struggle with fragmented data systems. CommonSpirit is⁣ working ⁢on a four-year plan to integrate patient records across its facilities, but achieving true interoperability remains ‍a challenge.

The use of synthetic data was discussed, but largely dismissed by health system representatives. Miller stated that CommonSpirit’s issue isn’t⁣ data‍ volume, but normalization and structure. She voiced concerns about validating AI-generated ‍insights ⁢in ‍clinical care.

Dr. Mitesh Rao, Founder & CEO of OMNY Health, cautioned ⁤about the risks of re-identifying anonymized data as AI and quantum computing advance.He explained that linking new components to de-identified data requires repeating the certification process to ensure patient‍ data remains protected.

panelists agreed that patients increasingly expect control over their health data. Rao argued that patients should have the ability to easily share their data, but current healthcare infrastructure makes this difficult.

Tonya Reeder, CIO at Walter Reed National Military Medical Center, emphasized the need for transparency in AI-driven clinical decision support. She stated ⁢that AI models must ⁢explain their reasoning,⁢ showing the data behind their recommendations to gain clinician⁤ trust.

Miller echoed these concerns, noting that AI must integrate seamlessly into clinical workflows without burdening providers. She emphasized that physicians need to understand why⁢ an‍ AI model suggests a particular diagnosis ⁤or treatment.

Panelists pushed back against fears that AI could replace human clinicians. Rao stated ‍that AI should augment decision-making, not automate it entirely, supporting human⁣ expertise.

Regarding budget concerns, Miller said AI investments must solve real problems.Reeder explained that Walter Reed focuses on AI use cases that directly improve patient outcomes, such as clinical decision support tools for diagnoses.

Rao acknowledged concerns about the energy consumption of AI models but pointed out that AI‍ is ‍becoming more efficient and open-source solutions are improving rapidly.

miller suggested that AI⁤ investments must align with ⁢overall digital strategies, emphasizing the need‍ for clean, structured, and standardized data ‍before building AI on top of it.

Panelists debated whether health systems should rely on external governing bodies to set AI policies or develop their own internal ⁤frameworks. Rao expressed⁢ skepticism about non-governmental agencies imposing regulations, while Miller argued that providers must take an active role in shaping AI standards.

Miller concluded that the industry’s excitement over AI must be balanced with practical concerns, emphasizing the need for a strong data foundation. Rao emphasized the need for adaptability ⁢and trust in data, ⁤models, and patients.

“If we build AI the right way-with transparency, integrity, ‍and a strong data⁢ foundation-it won’t just make ⁣us more efficient,” Miller ⁤concluded, “it will help us‍ deliver better ⁢care.”

What’s next

Healthcare organizations will continue to grapple with integrating AI into⁢ clinical workflows, focusing on data quality, transparency, and security to ensure patient trust and improve outcomes.