Steven Lin, MD
Unleashing the Power of AI in Primary Care
Steven Lin, MD
Unleashing the Power of AI in Primary Care
Steven Lin, MD, family physician and section chief of general primary care overseeing 150 clinicians, reached a tipping point as he witnessed the impact of overloading primary care physicians with too many administrative burdens.
“I was seeing rampant burnout,” Lin says. “Faculty were leaving us left and right.”
With the aim of revitalizing primary care, in 2019, Lin founded Stanford Healthcare AI Applied Research Team, or HEA₃RT. Its mission is threefold: accelerate the application of artificial intelligence (AI) and machine learning (ML) into the primary care space; support rigorous scientific AI implementation research; and address issues of diversity, equity, and inclusion in AI development.
HEA₃RT’s approach to fulfilling its ambitious mission is through aggressive collaboration with industry, academia, nonprofits, and government. Partnerships to date include projects with Google, Microsoft, and the National Academy of Medicine.
Automating Processes So Clinicians Can Spend More Time at the Bedside
Lin believed artificial intelligence and machine learning technologies could help alleviate physician burnout. Yet, despite half of all health care delivery occurring in primary care, only 3% of FDA-approved artificial intelligence and machine learning tools are actually built for it. Moreover, only a small fraction of the tools make it to production, and those that do seldom undergo rigorous evaluation.
Lin imagined that by automating burdensome parts of the clinical processes — clinical documentation and patient messaging, for example — it could free up enough space to allow primary care doctors to spend more time at the bedside, rekindling the patient-doctor relationship and allowing physicians to focus on the work they were trained to do.
“Providers are not worried about whether or not they can diagnose and treat patients,” Lin points out. “They’re worried about burning out and leaving medicine altogether because the amount of work they must do is unsustainable.”
A Bridge Between Data Scientists and the Front Lines of Health Care
Lin envisioned a transformative path for primary care, but the disconnect between AI tools designed in the lab and their actual implementation on front lines hindered progress.
To overcome these barriers, the team at HEA₃RT is composed not of data scientists but of quality improvement experts, implementation scientists, clinicians, and nurses.
“We serve as that bridge between the data science and operations world,” says Margaret Smith, HEA₃RT’s director of operations.
Amelia Sattler, MD, addresses Hea3rt Lab staff. From left: Timothy Tsai, DO; Yejin Jeong; Steven Lin, MD; Trevor Cromwell; Betsy Yang, MD.
Smith, who has a background in quality improvement implementation science, said that communication can get “messy” when navigating the different languages spoken by data scientists and operational healthcare experts.
By relying on people gifted in communication and collaboration, HEA₃RT is better positioned to propel the integration of AI solutions into the front line of health care. And by doing so, they are reinvigorating the spirit of primary care with energy-saving technologies.
“We’re well-versed in the operational language, and we’ve learned the technology language,” Smith notes. “We can help translate and bring those groups together.”
Steven Lin, MD (right), with Timothy Tsai and Hea3rt Lab staff
Steven Lin, MD, family physician and section chief of general primary care overseeing 150 clinicians, reached a tipping point as he witnessed the impact of overloading primary care physicians with too many administrative burdens.
“I was seeing rampant burnout,” Lin says. “Faculty were leaving us left and right.”
With the aim of revitalizing primary care, in 2019, Lin founded Stanford Healthcare AI Applied Research Team, or HEA₃RT. Its mission is threefold: accelerate the application of artificial intelligence (AI) and machine learning (ML) into the primary care space; support rigorous scientific AI implementation research; and address issues of diversity, equity, and inclusion in AI development.
HEA₃RT’s approach to fulfilling its ambitious mission is through aggressive collaboration with industry, academia, nonprofits, and government. Partnerships to date include projects with Google, Microsoft, and the National Academy of Medicine.
Automating Processes So Clinicians Can Spend More Time at the Bedside
Lin believed artificial intelligence and machine learning technologies could help alleviate physician burnout. Yet, despite half of all health care delivery occurring in primary care, only 3% of FDA-approved artificial intelligence and machine learning tools are actually built for it. Moreover, only a small fraction of the tools make it to production, and those that do seldom undergo rigorous evaluation.
Lin imagined that by automating burdensome parts of the clinical processes — clinical documentation and patient messaging, for example — it could free up enough space to allow primary care doctors to spend more time at the bedside, rekindling the patient-doctor relationship and allowing physicians to focus on the work they were trained to do.
“Providers are not worried about whether or not they can diagnose and treat patients,” Lin points out. “They’re worried about burning out and leaving medicine altogether because the amount of work they must do is unsustainable.”
Steven Lin, MD (right), with Timothy Tsai and Hea3rt Lab staff
A Bridge Between Data Scientists and the Front Lines of Health Care
Lin envisioned a transformative path for primary care, but the disconnect between AI tools designed in the lab and their actual implementation on front lines hindered progress.
To overcome these barriers, the team at HEA₃RT is composed not of data scientists but of quality improvement experts, implementation scientists, clinicians, and nurses.
“We serve as that bridge between the data science and operations world,” says Margaret Smith, HEA₃RT’s director of operations.
Smith, who has a background in quality improvement implementation science, said that communication can get “messy” when navigating the different languages spoken by data scientists and operational healthcare experts.
By relying on people gifted in communication and collaboration, HEA₃RT is better positioned to propel the integration of AI solutions into the front line of health care. And by doing so, they are reinvigorating the spirit of primary care with energy-saving technologies.
“We’re well-versed in the operational language, and we’ve learned the technology language,” Smith notes. “We can help translate and bring those groups together.”
Providers are not worried about whether or not they can diagnose and treat patients. They’re worried about burning out and leaving medicine altogether because the amount of work they must do is unsustainable.
Google, a Case Study for Success
HEA₃RT doesn’t stop at implementation. The team is dedicated to producing equity-driven health research around artificial intelligence by working with the biggest players in the technology space.
Their collaboration on the Google product DermAssist, an app equipped with advanced machine learning that diagnoses skin conditions from images and alerts users about the urgency of seeing a doctor, is a prime example of how they apply all three prongs — primary care, implementation research, and equity.
The app addresses the issue of limited access to dermatology care worldwide, particularly in rural areas. Primary care physicians handle 70% of skin cases, much more than dermatologists.
When it comes to issues of equity in AI, over the years, an outsize effort has been exerted upon addressing biased algorithms. While it’s important, Lin notes a whole other side of equity that includes involving patients and underserved communities in conversations about AI design and development.
Seeking HEA₃RT’s assistance, Google wanted research conducted to assess the app design and algorithm performance across diverse skin tones and use cases. A study conducted in partnership with Santa Clara Family Health Plan, serving a low-income community of mostly Latinx and Vietnamese individuals, provided valuable feedback and performance data. This collaboration advanced research and demonstrated that the app worked on different skin colors and included underrepresented populations.
Facilitating the collaboration with Google and Santa Clara Family Health Plan illustrated HEA₃RT’s commitment to rebuilding trust among underrepresented communities, Lin says.
Amelia Sattler, MD, addresses Hea3rt Lab staff. From left: Timothy Tsai, DO; Yejin Jeong; Steven Lin, MD; Trevor Cromwell; Betsy Yang, MD.
ChatGPT Accelerates Innovation
With a successful track record of collaboration, HEA₃RT isn’t afraid to partner on the latest cutting-edge technology. When ChatGPT’s consumer-friendly artificial intelligence program burst on the scene, it completely altered health care’s historically timid approach to adopting artificial intelligence.
“It has completely changed the AI/ML world to the point that every single health system is tripping over itself to incorporate it,” says Lin.
As the ChatGPT boom created a sense of renewed excitement and potential in the industry, HEA₃RT jumped in with both feet. The team is partnering with Stanford Medicine Technology and Digital Solutions to use ChatGPT to draft responses to patient messages, an incredibly burdensome task for primary care physicians.
“That project is not happening in the span of years — it’s happening in weeks,” Lin says. “ChatGPT is an example of how one particular, remarkable piece of technology has just taken the world by storm.”
Providers are not worried about whether or not they can diagnose and treat patients. They’re worried about burning out and leaving medicine altogether because the amount of work they must do is unsustainable.
Google, a Case Study for Success
HEA₃RT doesn’t stop at implementation. The team is dedicated to producing equity-driven health research around artificial intelligence by working with the biggest players in the technology space.
Their collaboration on the Google product DermAssist, an app equipped with advanced machine learning that diagnoses skin conditions from images and alerts users about the urgency of seeing a doctor, is a prime example of how they apply all three prongs — primary care, implementation research, and equity.
The app addresses the issue of limited access to dermatology care worldwide, particularly in rural areas. Primary care physicians handle 70% of skin cases, much more than dermatologists.
When it comes to issues of equity in AI, over the years, an outsize effort has been exerted upon addressing biased algorithms. While it’s important, Lin notes a whole other side of equity that includes involving patients and underserved communities in conversations about AI design and development.
Seeking HEA₃RT’s assistance, Google wanted research conducted to assess the app design and algorithm performance across diverse skin tones and use cases. A study conducted in partnership with Santa Clara Family Health Plan, serving a low-income community of mostly Latinx and Vietnamese individuals, provided valuable feedback and performance data. This collaboration advanced research and demonstrated that the app worked on different skin colors and included underrepresented populations.
Facilitating the collaboration with Google and Santa Clara Family Health Plan illustrated HEA₃RT’s commitment to rebuilding trust among underrepresented communities, Lin says.
ChatGPT Accelerates Innovation
With a successful track record of collaboration, HEA₃RT isn’t afraid to partner on the latest cutting-edge technology. When ChatGPT’s consumer-friendly artificial intelligence program burst on the scene, it completely altered health care’s historically timid approach to adopting artificial intelligence.
“It has completely changed the AI/ML world to the point that every single health system is tripping over itself to incorporate it,” says Lin.
As the ChatGPT boom created a sense of renewed excitement and potential in the industry, HEA₃RT jumped in with both feet. The team is partnering with Stanford Medicine Technology and Digital Solutions to use ChatGPT to draft responses to patient messages, an incredibly burdensome task for primary care physicians.
“That project is not happening in the span of years — it’s happening in weeks,” Lin says. “ChatGPT is an example of how one particular, remarkable piece of technology has just taken the world by storm.”