Doctors can now chat with patient records using new ChatEHR tool

Stanford Health Care has launched an AI tool that lets doctors, nurses, and other healthcare professionals query patient records with natural language, just like they would with consumer AI chatbots like ChatGPT. The new tool, ChatEHR, transforms how clinicians search through medical files, reducing the burden of documentation while improving their chances of catching critical details buried in patient records.

Saving time without missing details

Dr. Nigam Shah, Stanford Health Care’s chief data science officer who spearheaded the development, emphasized that “AI can augment the practice of physicians and other health care providers, but it’s not helpful unless it’s embedded in their workflow and the information the algorithm is using is in a medical context.”

ChatEHR is a major improvement over traditional electronic health record (EHR) systems and addresses a persistent challenge in modern healthcare: digital data overload. EHR systems often require users to navigate multiple tabs, search fields, and nested folders to access even basic information, contributing to frustrating delays in patient care.

Currently in its pilot phase, ChatEHR is being tested by a select group of 33 healthcare professionals at Stanford Hospital, including physicians, nurses, physician assistants, and nurse practitioners. These early adopters are evaluating the system’s performance while helping to refine its accuracy and expand its capabilities.

Dr. Sneha Jain, a clinical assistant professor of medicine and early user of the technology, noted that the tool helps clinicians avoid “scouring every nook and cranny” of medical records for needed information, allowing them to “spend time on what matters — talking to patients and figuring out what’s going on.”

Early feedback from pilot participants suggests the tool could save several minutes per patient interaction, time that adds up quickly over the course of a shift.

Fast answers in critical moments

The AI tool helps eliminate time-consuming tasks for doctors, like searching for test results or past procedures. When healthcare providers open ChatEHR, they see a simple chat interface where they can immediately start asking questions about patients: checking for allergies, reviewing recent test results, or inquiring about previous procedures and their outcomes.

Dr. Jonathan Chen, a hospital physician and assistant professor, highlighted the tool’s particular value in emergency situations. He explained that when patients arrive at the emergency room, doctors need to quickly understand “their whole story, what led up to this moment” including medications, side effects, surgeries, and their impacts.

The system also proves valuable for complex transfer cases, where patients arrive with extensive medical documentation that can span hundreds of pages. Rather than manually reviewing all materials, clinicians can use ChatEHR to generate relevant summaries and ask targeted follow-up questions.

Automations speed up evaluations

Beyond simple information retrieval, the development team is creating sophisticated “automations” that can perform evaluative tasks based on patient records. One example helps identify which patients may be eligible for transfer to Stanford Medicine-affiliated Sequoia Hospital, which offers additional patient room capacity.

Shah explained that these automated evaluations “save us the administrative burden of sifting through patient information and helps us quickly determine if a patient can be transferred, opening access to care here at Stanford Hospital.”

Additional automations under development include tools for assessing hospice care eligibility and identifying patients who may need enhanced post-surgical monitoring.

The developers stress that ChatEHR just helps find information and doesn’t give medical advice. Doctors still make all the decisions. The system pulls information exclusively from patients’ existing medical records, ensuring relevance and context.

To maintain the tool’s accuracy and reliability, the team is utilizing MedHELM, an open-source framework specifically designed for evaluating large language models in medical settings. Future updates will include citation features that show clinicians exactly where information originated within medical records.

Ready to (responsibly) roll

ChatEHR’s development began in 2023 when Stanford Medicine researchers, led by Shah and Anurang Revri, vice president and chief enterprise architect for Stanford Health Care’s Technology and Digital Services, recognized the potential applications of large language models in healthcare settings.

The project received support from Stanford’s Department of Medicine and the Center for Biomedical Informatics Research, demonstrating the institution’s commitment to improving patient care with new technology.

Dr. Michael Pfeffer, Stanford Health Care’s chief information and digital officer, described ChatEHR as “a unique instance of integrating LLM capabilities directly into clinicians’ practice and workflow,” expressing enthusiasm about bringing the technology to Stanford Health Care’s broader workforce.

The development team plans to gradually expand access to all clinicians who work with patient charts, following what Shah describes as responsible AI guidelines that prioritize accuracy, performance, and adequate educational support for users.

With doctors increasingly burdened by the demands of paperwork and data entry, AI tools like ChatEHR are poised to help them spend less time in front of computers and more time treating patients.