Stanford Medicine: Using AI to make healthcare more human-centred
Artificial intelligence presents a unique opportunity to advance health and medicine globally – from diagnosing diseases sooner to accelerating research to personalizing patient treatments.
The potential is immense, as are the potential risks without proper oversight. It’s critical that this new era is navigated responsibly, with regulations and safeguards to ensure ethical AI use which benefits both patients and healthcare providers.
At the crossroads of these emerging opportunities and challenges is Stanford Medicine – an integrated academic health system located in the heart of Silicon Valley in California, USA. Comprised of the Stanford School of Medicine, Stanford Health Care, and Stanford Medicine Children’s Health, Stanford Medicine’s mission is to improve human health through discovery-led research, training tomorrow’s health leaders, and delivering world-class patient care.
Since the 1960s, Stanford has been a pioneering force in artificial intelligence. This rich legacy continues today, particularly at the intersection of AI and medicine. By working across disciplines – ranging from data science to biomedical ethics – and collaborating with industry leaders and peers, Stanford Medicine’s vision is to chart the future of AI-enabled medicine, ensuring that tomorrow’s innovations are impactful, responsible, and equitable.
“We expect AI will transform patient care in ways we’ve only dreamt of,” said Priya Singh, Chief Strategy Officer and Senior Associate Dean at Stanford Medicine. “At Stanford Medicine, we’re committed to leading the way in applying this powerful technology to improve health globally while ensuring its principled development and use.”
Enhancing patient care
Today, multidisciplinary teams at Stanford Medicine are actively creating, testing, and iterating on AI tools to support and enhance patient care. Within Stanford Health Care, the adult healthcare delivery system, the use of AI spans roughly 30 different applications, from the automatic analysis of cardiac MRI images to the use of scoring systems that alert care teams about patients whose health status is at risk of deteriorating.
From these experiences, AI has proven to be a powerful tool to generate on-demand evidence when literature or guidelines are scant. One example of this is a service called the Clinical Informatics Consult Service. Developed over several years at Stanford Medicine, the service sifts through Stanford’s vast clinical data warehouse to provide physicians with details from past patient records, informing care plans for their current patients. Now available through Atropos Health, a company that licenses the technology from Stanford, the service has been used more than 1,000 times by our clinicians over the past year.
Another example is a tool developed at Stanford Medicine that can predict from an initial primary care visit which patients are likely to have undiagnosed peripheral artery disease, where narrowed arteries reduce blood flow to a patient’s arms and legs. Such insights can help clinicians deliver precise and preventive care to individual patients – informed by medical knowledge gleaned from entire populations and thousands of previous clinical interactions.
Stanford Medicine is also exploring the use of AI to ease some of the administrative burdens on clinicians that are known to erode their wellbeing and eat into time spent with patients. As one example, Stanford Medicine has piloted a tool to help clinicians draft responses to patient email inquiries, which often take hours to resolve after clinic each day. Using the service, clinicians review and edit AI-generated responses before the messages are sent. A study of the model’s pilot phase yielded promising findings, with clinicians reporting that it reduced their clerical burden and feelings of burnout. Another AI system at Stanford Medicine uses ambient voice recognition to securely listen to patient encounters and generate draft clinical notes for clinicians to review and finalize after the appointment.
“As an industry, we are only beginning to understand the many ways that AI will be able to improve the clinical experience for both our patients and our providers,” said Michael Pfeffer, MD, Chief Information and Digital Officer at Stanford Health Care and Associate Dean at the School of Medicine. “Implemented responsibly, these technologies have the potential to positively impact everything from how we train our surgeons to how we monitor our patients’ conditions in our hospitals.”
A high bar for responsible AI
Because healthcare is a uniquely sensitive setting, any AI must meet high standards for fairness, safety, usefulness, reliability, and efficacy. At a health-system level, Stanford Medicine has created a rigorous, multi-disciplinary process called the “FURM assessment” for evaluating AI models. At its core, this lifecycle review of AI encompasses a comprehensive assessment of the technology’s ethical rigor, safety, financial sustainability, and implementation feasibility – among other considerations. The overarching goal is to ensure that AI used by Stanford Medicine continuously adds value for patients, care providers, and the community.
The key to responsible AI is acting collectively. While individual organizations can pursue efforts to create standards for sound AI use and development, these standards must be shared – with clear accountability – among healthcare organizations, technology developers, and other key stakeholder groups. To this end, Stanford Medicine is collaborating with experts at Stanford University and across sectors to lead this important conversation and facilitate the adoption of shared responsible AI practices.
In 2023, Stanford Medicine joined with the Stanford Institute for Human-Centered Artificial Intelligence (HAI) to launch RAISE Health. This initiative was created to guide the responsible use of AI across biomedical research, education, and patient care by providing a go-to platform for resources, convening thought leaders and other stakeholders, and defining frameworks for ethical health AI. Stanford Medicine is also a founding member of the Coalition for Health AI (CHAI), a group of health systems, government agencies, and private companies working together to develop guidelines and guardrails for AI in healthcare.
“Without a doubt, now is the time to bring together top experts in technology, medicine, and policy to steer AI’s future, ensuring that this technology aligns with our goals and values,” said Sylvia Plevritis, PhD, professor and chair of biomedical data science at Stanford Medicine, and co-lead of RAISE Health’s faculty research council. “The scope of this challenge is bigger than any single entity, but with our collaborative DNA and multidisciplinary expertise, I am confident that Stanford Medicine is uniquely equipped to lead the way in this vital work.”
