As hospitals and health systems push to reduce friction in clinical workflows, EHR optimization has become mission-critical.
Seven executives at the nation’s largest health systems told Becker’s they need systems that support real-time clinical decision-making without disrupting workflows. Tools like ambient listening, automated summaries and clinical decision support rank high on their lists.
Question: How are you evaluating EHR vendors’ ability to support predictive analytics, AI integration, and care coordination in the next three to five years?
Editor’s note: Responses have been lightly edited for clarity and length.
Crystal Broj. Chief Digital Transformation Officer at MUSC Health (Charleston, S.C.): At MUSC Health, we adopt a platform-first approach when evaluating EHR vendors. We seek systems that are safe, reliable and capable of evolving with healthcare innovations to better meet the needs of our consumers without overwhelming our care teams. This means:
Core capabilities must keep pace: Vendors must demonstrate that their platform is advancing in key areas such as predictive analytics, AI-driven workflows and care coordination tools. We are particularly interested in features that reduce friction for our clinical and operational teams while improving patient outcomes.
Openness to integration: No platform can excel in every area. If a vendor’s native capabilities are not keeping up with market demands, we expect their architecture to allow for seamless integration with partner vendors. Important elements include APIs, FHIR-based interoperability and flexible data access to help us meet our needs.
AI and predictive insights in workflow: Predictive insights and AI tools should be integrated into daily operations rather than being confined to dashboards. We evaluate whether the EHR can intelligently surface insights and trigger actions in real time, directly within the clinical workflow.
Care coordination at scale: We assess whether the platform supports care coordination across different settings, featuring shared care plans, intelligent task management, and communication tools that function across teams and care environments.
Ultimately, we are not just purchasing EHR software; we are investing in a digital foundation. This foundation must be flexible enough to grow with us, open enough to allow best-in-class tools, and robust enough to support safe, scalable care delivery for the next generation of healthcare.
Nicole Gitney-Fahey, RN. Vice President and Chief Nursing Informatics Officer at BayCare Health System (Clearwater, Fla.): We are partnering with our EHR vendor to develop and implement AI that lives innately in the EHR. As we evaluate other AI vendors, we assess their ability to integrate into our EHR and create cohesive workflows for coordination of patient care.
Emily Jacobsen. Vice President of Clinical Systems and Chief of Clinical Informatics at University of Maryland Medical System (Baltimore): At the University of Maryland Medical System, our approach to evaluating EHR vendors is deeply rooted in the day-to-day realities of our clinicians. As both the VP of Clinical Systems and Chief of Clinical Informatics — and as someone who has spent years listening to the frustrations and hopes of our providers — I believe that any technology we adopt must first and foremost ease the cognitive and administrative burdens our teams face.
EHRs were meant to help, but too often they’ve become a source of burnout. We’re now looking for tools that do more than just store data — they need to intelligently highlight what matters, summarize complex information quickly and support clinical decisions in real time. Predictive analytics and AI should not be buzzwords; they should be practical, safe and equitable tools that help our caregivers spend more time with patients and less time clicking through screens.
Ultimately, we’re evaluating these technologies through a human lens: Do they help our clinicians focus on what truly requires their expertise? Do they reduce harm and improve care coordination? And do they do so in a way that respects the diversity of our patients and providers? If the answer is yes, then we’re moving in the right direction.
Frank Liao, PhD. Senior Director Digital Health and Emerging Technologies at UW Health (Madison, Wis.): At UW Health, our approach to evaluating EHR vendors’ capabilities in predictive analytics, AI integration, and care coordination is rooted in best practices that balance clinical and operational benefits, safety, technology architecture and long-term value.
For example, with safety, we have committees and frameworks providing safety oversight and ongoing recommendations. We maintain active and strong partnerships with our vendors to stay current with their long-term vision, and we incorporate their near-term product roadmaps into our own planning and prioritization processes. With system architecture, we prioritize systems integration and seamless workflows, while balancing speed to value. We feel strongly that these capabilities must ultimately improve the care that we provide our patients and communities and benefit our workforce over the next three to five years.
Lee H. Schwamm, MD. Senior Vice President and Chief Digital Health Officer at Yale New Haven (Conn.) Health: As an Epic customer of many years, Yale New Haven Health has been working closely with them over the past decade to promote the development and deployment of AI tools within the platform. We partnered early with Epic to perform local validation of their AI sepsis prediction algorithms. We were one of the first AMCs to join Cosmos to expand the AI research portfolio for our clinical research community and deploy Abridge for ambient clinical documentation deeply integrated within Epic. Yale New Haven Health is a founding partner of CHAI, and we are about to release our own operating model for healthcare AI implementation which includes a premarket assurance assessment complete with model card requirement, a local silent validation phase, and a post-deployment monitoring plan for all AI products. For our Epic implementation, we will be evaluating them as both a platform for implementation of other AI vendor products as well as an AI vendor themselves. These criteria will be applied in a risk-stratified manner based on a comprehensive risk assessment that includes financial, reputational, statistical and clinical dimensions. These processes are critical to ensuring a fair and responsible implementation of AI to ensure continued public trust.
Nirav Shah, MD. Associate Chief Medical Informatics Officer of AI and Innovation at Endeavor Health (Evanston, Ill.): Our AI framework can be distilled into three components: strategize, trust and realize. Our evaluation includes understanding whether the solutions being offered align with our strategic priorities around AI deployment, which includes intelligent clinical decision support, early disease detection, optimizing the consumer and human experience, optimizing resource allocation and automating administrative tasks. And within those strategies, we need to make sure that we are solving key problems for our team members and not implementing technology that looks for a problem. For trust we assess the vendor’s approach to ethical, explainable and responsible AI — ensuring transparency, clinical oversight for decision-making, and out-of-the-box tools to validate and monitor these tools efficiently and effectively. Finally, our evaluation focuses on solutions that allow us to realize value by demonstrating measurable improvements in outcomes that align with the quintuple aim.
Sudipto Srivastava. Chief Data and Analytics Officer at Montefiore Einstein (New York City): At Montefiore Einstein, we have been using artificial intelligence and machine learning tools for early disease detection and to develop personalized risk scores for certain patients’ readiness for treatments and potential complications. These algorithms have been particularly valuable with life-threatening conditions like sepsis which can progress rapidly and be unpredictable in its progression. Results include Montefiore clinicians seeing red flags for admitted patients at elevated risk of mortality three to five days in advance. We also developed a risk score to predict the onset of sepsis six hours in advance, enabling our doctors to have one of the most precious tools to improve health outcomes — time.
These early successes have sparked demand among our clinical and operations teams for predictive analytics and care coordination. As we look to build on these efforts, we’re focused on the tools that allow us to get back to our patients faster and further aid clinical productivity. This is where ambient listening comes in. We need tools to reduce the burden of documentation for our clinical teams. Furthermore, we’re attempting to address how to handle the vast amount of AI innovation that is happening outside of a classic EHR setting — like in the radiology space for stroke detection, breast cancer analysis, detecting fractures, etc., or the use of agentic AI in the finance space within revenue cycle management. Over time, most solutions will be available through our EHR, and our EHR vendor has been very proactive in introducing many options for us, but we need to be open to experimenting with multiple partners — like with the startup community, our vendor partners and research partners — to enable academic health systems like ours to pivot fast as industry and research evolve.
We follow a process that has strong guardrails, and all our predictive/AI solutions go through our AI clinical advisory group, an AI governance team, tech evaluation and financial feasibility. Our last evaluation step, which is the simplest but sometimes, as an industry, not prioritized, is that “old-fashioned” phone call. By speaking with our colleagues at other healthcare systems, we can filter out hype-free dialogue and get down to what works.
The post 7 leaders on how EHRs must evolve by 2030 appeared first on Becker’s Hospital Review | Healthcare News & Analysis.
Health IT
