AI ⁢in Healthcare: Cutting⁤ Through the‍ Hype to Drive Real Impact

‍ Updated June 22, 2025
‍

At ViVe 2025, healthcare⁤ IT leaders​ explored how artificial intelligence (AI) can transform clinical workflows, reduce clinician burden, and improve patient outcomes. ⁢Panelists from Cleveland ‍Clinic‌ and Stanford Medicine agreed that the true value of AI in healthcare lies in careful implementation and thorough evaluation.

Dr. Rohit Chandra, executive ⁢vice president and⁤ chief‍ digital officer at cleveland Clinic, and Dr. Michael Pfeffer, chief details and digital officer at Stanford Medicine, shared‍ insights on their AI ⁤initiatives. Both‌ emphasized that ‌AI’s success should be measured by‍ tangible clinical and operational results, not just adoption for innovation’s sake. The discussion highlighted that AI adoption in healthcare is still developing.‌ Some applications ‌show promise, while ⁤others​ remain experimental.

AI Scribes Ease Burden

One immediate application of AI ⁢in healthcare is AI-powered scribes for physician documentation. These ‍scribes aim⁣ to reduce administrative tasks, ‌allowing doctors to focus on‌ patient care. Chandra noted that documentation requirements⁢ contribute to physician burnout, making AI-driven​ automation a priority.”Burnout is ⁤a huge issue,”‍ he said. “reducing documentation‍ time can improve job⁢ satisfaction and⁢ patient care.”

However, AI scribe ​success involves more than just time savings.Pfeffer explained ​that Stanford Medicine ⁢evaluates ⁤AI implementations using a framework called Fair, Useful, Reliable Models (FIRM).This framework measures clinician burnout, turnover, and efficiency, rather than just tracking time saved on documentation.”If‌ we turn off an AI tool ⁣and ​get​ hundreds‍ of angry​ emails from clinicians, we know it’s ​working,” Pfeffer said. “That’s rare in healthcare ​IT.”

Panelists also discussed AI’s role in⁤ clinical decision-making, particularly in identifying conditions ‍like⁣ sepsis.⁢ Cleveland⁢ Clinic is developing AI-driven sepsis detection to improve early intervention. Chandra​ explained that‍ current sepsis detection workflows rely⁢ on alert-based systems ‍that can⁣ cause clinician fatigue. “Many sepsis alert systems today create​ fatigue rather than delivering real therapeutic benefit,” he⁢ said.⁣ “Our ‌AI model optimizes detection,reducing unneeded‌ alerts while improving early intervention rates.”

Setting Clear⁢ Expectations for AI

pfeffer stressed ⁤the importance of​ setting clear expectations for AI’s⁤ impact ⁤on clinical ‌decisions. “Every ⁤AI deployment should have a clear outcome measure,”⁤ he said. “If it doesn’t improve⁤ care, it shouldn’t be ​in use.” He ⁢added ⁣that a key challenge is ensuring AI doesn’t simply‌ shift responsibility‌ to already overburdened clinicians. “If⁤ AI is supposed to increase efficiency but⁢ still requires manual oversight, it’s not a true solution,”⁢ he noted. “Human-in-the-loop AI sounds good in theory, ‌but if clinicians are expected to verify every⁢ AI-generated proposal, we’re just adding another layer of⁢ work.”

As AI ⁤adoption increases,‍ health systems should be wary of vendor claims ​and hype.Chandra emphasized openness ‍in implementation. “Never tell a CIO your product ‘integrates easily,'” ⁤he warned.⁢ “Every‌ system requires ​work to implement.” He noted ⁣that health systems ⁢have tight budgets and limited IT resources, making it ‍vital to prioritize AI tools that ⁢deliver measurable value. Pfeffer agreed, ⁢adding that health systems should evaluate AI⁣ tools using a ‍balanced approach, recognizing that ⁣not all applications ⁣will‌ yield immediate financial returns.

“Value isn’t always about ⁣money,”⁤ Pfeffer explained. “Some ‍AI ​tools improve patient outcomes and reduce clinician burnout, which indirectly ‌benefits the health ​system.” Chandra ⁢emphasized that AI success requires structured evaluation and⁢ a willingness to adjust or abandon projects that‍ don’t deliver results. “We don’t expect⁣ perfection,” he ⁢said. “But if an AI‍ tool isn’t delivering real clinical or ‌operational improvements,we⁣ turn it ‌off.”

Another⁣ discussion point was AI’s role in ‌improving patient monitoring and chronic disease management. Chandra noted that Cleveland Clinic is exploring home-based care models that use AI to extend hospital-level⁢ monitoring to patients at home. “We’re expanding a program where we⁣ can literally send patients a set of monitoring devices and‌ have them cared for remotely. This allows us to provide​ hospital-level care while ‌freeing up capacity for the most critical patients.”

The full potential of AI in remote⁣ monitoring remains ‍untapped,⁤ but​ panelists agreed it offers a critical possibility to improve long-term‌ patient care. Chronic disease management could benefit from AI-powered analytics that ⁢provide continuous insights ​rather than ​relying on episodic physician visits. “The future of healthcare‌ isn’t just about point-of-care interactions,” Chandra said. “It’s about ⁤continuous care, where AI helps​ manage a patient’s condition in real-time rather than ‍waiting⁣ for their ‌next scheduled appointment.”

Pfeffer added that AI is also⁣ set to transform‍ how health ‌systems handle‍ data analytics.‍ Traditionally, clinical ​data analysis ⁢has been retrospective, ‍but AI enables real-time​ insights that can improve decision-making. “We’ve done some⁢ work at Stanford where we can ask real-time ‌questions ⁣of our EHR data and get meaningful insights. Instead of running retrospective reports, we can use‍ AI models ⁣to interpret complex data sets and provide actionable intelligence ⁣in real time.”

despite these⁤ promising ⁤applications, panelists cautioned about‍ AI’s limitations. ‍Chandra ‌expressed skepticism about the readiness of AI-powered clinical decision​ support tools, particularly in frontline care settings.”AI⁤ in its ⁤current state isn’t ready to take over frontline clinical decisions,” he said.

“The safety‍ protocols and governance structures aren’t fully in place ‍yet.” Pfeffer agreed, noting ‍that ⁣while AI has made critically importent progress in imaging and predictive analytics, its integration into direct patient care‌ still requires⁢ careful ⁢oversight.

Looking Ahead in AI for‍ Healthcare

Panelists emphasized the need⁤ for rigorous evaluation and ‌thoughtful deployment strategies to ensure AI‌ delivers⁣ real value in healthcare. “Technology moves fast, but healthcare moves slowly,” Chandra said. “The key is to⁤ bridge that gap-experiment responsibly, measure impact, and ensure ⁤AI truly improves patient​ care.”

As the healthcare industry continues to navigate AI’s‍ rapid evolution, panelists encouraged IT executives to remain focused ⁢on ​outcomes rather than ⁤hype. Pfeffer added: “we should always ​ask, ​’What problem ⁣are we solving?’ If we can’t clearly articulate the value AI brings to patient care, clinician‌ well-being, or operational efficiency, ⁣then we need to‍ rethink our approach.”

What’s​ next

Healthcare leaders must focus ⁤on building AI ‍solutions that genuinely enhance patient⁤ care and clinician​ efficiency. ‌Pfeffer⁢ said:⁤ “AI isn’t just⁤ about automation-it’s about creating a healthcare system that works ⁤better for ⁤both patients and providers.”