HLTH 2026: Elsevier upgrades ClinicalKey AI with full patient EHR context

A doctor using a futuristic computer, symbolizing AI
San Juan Regional Medical Center in Farmington, New Mexico, is using ClinicalKey AI’s new patient context functionality to provide quality care with fewer resources. (ismagilov/GettyImages)

LAS VEGAS—Elsevier unveiled this week a major upgrade to ClinicalKey AI, enabling the AI-powered clinical decision support tool to access a patient’s full electronic health record and evaluate treatment plans against the company’s evidence-based medical library.

As AI advances, Elsevier has strategically shifted from a traditional academic publisher— it’s one of the world's largest publishers of scientific, technical and medical content—to an AI-driven analytics information company, incorporating generative and agentic AI into its healthcare and academic research.

ClinicalKey AI is a HIPAA-ready point-of-care decision support tool that draws answers from full-text articles from approximately 1,000 world-leading journals. Elsevier’s upgraded ClinicalKey AI is already deployed at more than 1,200 hospitals, and the company says the solution integrates with all electronic health record providers. Elsevier recently struck a content agreement with the JAMA Network that brings full-text content and visuals from the JAMA Network Family of Journals to ClinicalKey AI.

The company has rapidly expanded its AI capabilities and features, including ClinicalKey Nursing AI, an AI-powered tool built specifically for nurses and nursing workflows. Elsevier has also developed specialized AI assistants, such as LeapSpace, an AI-assisted workspace for researchers that’s built on 15+ million peer-reviewed articles and book content and 100 million Scopus records, according to the company. 

Three months ago, the company rolled out its "Ask AI" feature across more than 1,500 U.S. hospital sites using its flagship ClinicalKey platform. With the generative AI feature, clinicians can ask conversational clinical questions and receive direct, cited answers drawn from evidence-based sources, including premium journals and content from leading medical organizations. Unlike general-purpose chatbots, Ask AI does not pull from the open web and does not store, process or retain patient-identifiable information, according to the company.

In June, the company acquired Wellsheet, a health technology company that aggregates and synthesizes patient chart data—including notes, labs, imaging and medications orders—from EHRs to provide relevant clinical intelligence at the bedside. The integration of WellSheet's patient-chart analysis enhances Elsevier’s clinical decision support with context-aware AI, according to executives.

“The problem we’re trying to solve here is that physicians are challenged by so many things when they’re taking care of patients; top among them is patient complexity. Patients have more and more conditions and they’re taking more and more medications. It can be hard to really understand the full picture and determine how to proceed,” Dr. Claudine Lott, physician executive for clinical solutions at Elsevier, told Fierce Healthcare during the HLTH conference this week.

“Our original ClinicalKey AI tool has been a place where the clinician can go and they can ask a question and it searches this curated database of evidence-based content we’ve given it and gives them an answer. But there was a bit of a missing piece in that it wasn’t capturing that patient context. If you wanted to get an answer that’s really tailored to a patient, it was limited to what you can type in,” she said.

She added, “WellSheet has 10 years of experience with looking at the entirety of the patient chart and helping to surface that to the clinician.”

There is a growing market for AI-powered clinical decision support products, but Elsevier contends its advantage lies in its depth of patient context and its large, curated content base, with a wholly owned, large proprietary library of peer-reviewed STM literature, including content from millions of articles, journals and books.

San Juan Regional Medical Center in Farmington, New Mexico, is using ClinicalKey AI’s new patient context functionality to provide quality care with fewer resources. With smaller clinical and quality teams than large academic health systems, and in a state where nearly 40% of residents rely on Medicaid, the medical center, which serves the Four Corners area, has found this support especially valuable, according to Carlo Hallak, M.D., physician executive, information services, at San Juan Regional Medical Center.

A non-profit, community-owned acute care hospital and Level III trauma center, San Juan Regional Medical Center operates outpatient and specialty clinics alongside its main hospital campus, with associated facilities such as San Juan Regional Rehabilitation Hospital and San Juan Health Partners clinics.

Hallak emphasized that Elsevier is a primary source of medical knowledge rather than a “wrapper” for other AI platforms, and that was a key distinction for the organization, Hallak noted. And ClinicalKey AI is designed for the entire care team. “The hospital is run by nurses, mid-level providers, other caregiver teams,” he told Fierce Healthcare at the HLTH conference.

The medical center aims to move clinicians to enterprise LLMs integrated with the electronic health record that can provide the full patient context, Hallak said.

"We're already seeing how ClinicalKey AI can be instrumental at the bedside and help improve care for our patients. As a rural healthcare center, we are proud to say that by using ClinicalKey AI to check each patient’s care against the same evidence base that major health systems rely on, we’re able to deliver the standard of care patients expect in any major city or leading academic center," Hallak said.

He added, "ClinicalKey AI reads the whole chart, checks it against the evidence and shows exactly where each finding comes from. That support is invaluable for a hospital like ours, and we’re already seeing strong adoption of ClinicalKey AI among our clinicians."

Hallak cited an example of a physician who used ClinicalKey AI to correctly identify viral meningitis, preventing an aggressive and unnecessary treatment course for bacterial meningitis. 

ClinicalKey AI supports the physician’s judgment rather than replacing it, helping the clinician get to relevant evidence faster, he said, noting that, in the case of the viral meningitis patient, the physician’s clinical judgment was critical in recognizing that something was unusual and then consulting colleagues. The AI tool then helped accelerate the next step by quickly surfacing information the clinician otherwise would have found by searching the medical literature.

Physicians also use the tool during weekly complex case discussions to quickly review trending medications, images and test results in one summarized sheet, he noted.

“It’s going to expedite a lot of the work they do when it comes to these complex cases,” he noted.

The AI solution can quickly summarize a complex patient history at the point of care, and this can help care teams in the emergency department understand a frequent ED patient, Hallak said.

One patient has visited their emergency department 95 times in less than a year, in addition to making numerous visits to neighboring health systems, Hallak said. When that patient arrives, clinicians who aren't familiar with him would traditionally have to dig through the chart to understand his history. With Elsevier's ClinicalKey AI tool, they can ask for a summary of the patient's recent visits and care and get that information within seconds, he noted.

Because patients who frequently visit hospital EDs typically receive care at multiple organizations, San Juan Regional Medical Center is also working with other local health systems on interoperability efforts so clinicians can get a more complete picture of what happened during the patient's visits across different facilities.

As a rural hospital with staffing constraints, technology is essential to filling workforce gaps and supporting care delivery, Hallak contends.

"We don't have enough physicians, enough nurses, enough transporters, enough caregivers. We will never be able to have enough of those," he said.

Hallak envisions ClinicalKey AI will be particularly valuable to discharge patients from the hospital faster. 

Generative AI tools can help speed up patient discharges by identifying barriers, assessing patients’ care needs and surfacing relevant information from both structured and unstructured data. Because the platform is designed for the broader care team, case managers and physicians can access the same information, while push notifications can alert clinicians when test results are available so they can act quickly, Hallak said.

"From an operational perspective, it allows for this care team coordination of care, and also from a physician perspective, it provides the context whenever there's something hanging and they're trying to understand more what's happening with the patient discharge. You could be a hospitalist, and you've been off shift for like seven days, and then you come to the hospital and now the patient is on their tenth day, but you inherit that discharge. Going through, reading every progress note for every single day, reading every consult for every subspecialty that looked at this patient usually would take two to three days to actually get the patient out of the building safely," he said.

Rather than manually compiling information for a discharge summary, clinicians can use ClinicalKey AI to aggregate relevant information from across the patient’s record, trace it back to its source and review the AI-generated summary before signing off on the discharge, according to Hallak.

"It is a game changer, in my opinion. One of our hospitals said it used to take me three days to discharge a patient; now it takes me less than three hours," he said.

Going forward, the system expects to measure operational outcomes such as length of stay, safe discharge and multidisciplinary-rounds use, according to Hallak. The medical center also plans to evaluate clinicians' use of the tool and how it changes care decisions.

The medical center has also deployed other AI tools such as Oracle Health's clinical AI agent for ambient documentation.

"These things are crucial. We're hoping to bring in a clinical AI agent for nurses as well to help the whole care team be able to eliminate the noise, the friction that typically EHRs or IT would bring to the conversation," Hallak said.

The medical center is taking a risk-based approach to AI governance, using source verification and peer review for AI-generated summaries while applying greater scrutiny to tools that are predictive or suggestive, Hallak noted. That oversight involves clinical experts as well as compliance, legal and cybersecurity teams along with traditional vendor security and product-vetting requirements.

AI governance should become stricter as the technology moves from summarizing information to recommending or taking actions, he said.

Elsevier designed ClinicalKey AI with security and trust in mind, including HIPAA compliance, citations that show where information comes from and safeguards that keep the tool from prescribing what clinicians should do. Instead, it surfaces relevant information to help clinicians make their own decisions, Lott said.

"Health systems tell us two things stand in the way of adopting AI: tools that don’t work with their existing EHR partners, and concerns about security, compliance and responsible use," Omry Bigger, president of Clinical Solutions at Elsevier, said. "ClinicalKey AI was designed to address both. As an EHR-supported and HIPAA-compliant ready solution, ClinicalKey AI integrates seamlessly across health systems to provide clinicians with trusted evidence and patient context that gives them tailored clinical insights for each patient at the point of care."