Care Redesign

What AI Means for Doctors and Doctoring

Interview · September 3, 2019

Nirav Shah and Tom Lee head shots

Thomas H. Lee, MD, interviews Nirav R. Shah, MD, MPH, Senior Scholar for the Clinical Excellence Research Center at Stanford University.


Tom Lee:  This is Tom Lee for NEJM Catalyst. Today we’re talking with Nirav Shah, a Senior Scholar at Stanford’s Clinical Excellence Research Center. Nirav has one of the most interesting paths in health care that I know and he’s also one of the sharpest thinkers I’ve seen in medicine. Today, we’ll be talking about a narrow little question: the future of health care and the role of the physician in it.

In all seriousness, Nirav has a set of thoughts on this topic that are a combination of insightful, humble, and even moving. I’m sure you’re going to enjoy them, too. I first started hearing about Nirav when he became Commissioner of the New York State Department of Health at a ridiculously young age. How old were you at the time, Nirav?

Nirav Shah:  Thirty-eight.

Lee:  Thirty-eight years old. And then after that he became Chief Operating Officer for Clinical Operations at Kaiser Permanente Southern California, one of the biggest parts of Kaiser Permanente and one of the most forward-looking places. He was the clinical creative force, aligning with Ben Chu in some golden years for KP. Now he’s at Stanford, working at one of the most forward-looking care redesign groups in the country.

He’s an internist who understands health care delivery’s strengths and limitations and its potential future. Let’s start with what’s looming in AI [artificial intelligence], mainly to set the context for talking about what it means for doctors and doctoring. I’ll start with a superficial question: Are you bullish on AI?

Shah:  Thanks, Tom. I am bullish on AI in health care, though I’m a little bit afraid as well. The story of new technology in health care, unlike in most other industries, is that innovation equals higher cost, not increased efficiency. The reasons for this are complex, but we are desperately in need of technology and solutions that can lower the cost of great care, that can improve quality, and that can also decrease the burden on clinicians. We don’t need EHRs [electronic health records] to cause massive burnout again. Broadly speaking, AI can help us make progress toward the Triple Aim, but we have to be honest about its limitations, its potential for bias, its privacy implications, and so on.

Lee:  A lot of practicing doctors like me have hopes that AI will do something great, but we also hope that it will do those great things after we’re retired or dead because we are overwhelmed with new stuff already. How do you look at the situation of practicing doctors regarding AI, and what advice do you have for them? That’s really the meat of our discussion today.

Shah:  AI is not going to solve all the problems in health care despite what some people might say. How I think about it, and how I hope it will unfold, is that it will take away some of the boring stuff, some of the easy stuff first, leaving behind some of the more important and interesting stuff for doctors to do.

One of the great things about a physician is not only knowing the right thing to do in a complex set of circumstances, but the ability to act on that knowledge and to do what’s right. AI can help us hopefully close that knowing-doing gap, and so there’s hope we don’t have to spend as much time documenting, charting, billing, as much as getting time back to spend time caring, empathizing, and listening to our patients.

Lee:  You make a good case that AI is going to do a lot of stuff that frankly doesn’t take a tremendous amount of human intelligence but takes a lot of time for us to do, but we are going to have to change, too, in order to effectively work with AI. What are the kinds of things we’re going to have to learn? I want to set the stage for our clinicians to get ready to plunge in and recognize what they need to learn, so that we can learn it and begin to change in the way that will make us more effective.

Shah:  One way to think about AI would be how we’re already starting to think about team-based care. In the past, the doctor was someone who worked alone, who worked and was the head of a practice that usually consisted of one or two physicians. Now we realize that higher-quality care can be delivered by teams. If you think of AI as a member of the team, as a thoughtful, intelligent member of the team that can do some things very well and other things not very well, how do we leverage that new partner in the team?

If you think about it that way, it’s perhaps not as scary as yet another foreign invasion into our practice and forcing us to change — but there will be changes needed. Some of the changes that are needed are in how we actually do the work of care, how we practice medicine.

If you think of a team that extends beyond your clinic into the patient’s home, into the community, there are ways we can nudge people along into better care, leveraging AI and other tools in ways that we alone as a physician, as a practice — including a nurse practitioner, maybe a social worker or a community health worker, and other ancillary personnel — cannot do alone. That is the potential.

In the short term, for the next 3 to 5 years, I don’t see many disruptions in how we care for practice, but over the longer term and the middle term, we have to start to think about how we can best leverage this tool just as we’ve leveraged electronic health records to create population health dashboards, [or] how we’ve started doing daily team huddles around what the gaps are that we need to close around a given patient’s care for the rest of the morning. Those are the kinds of things that we have to start to think about.

Lee:  This is an interesting angle you’re introducing here that frankly I haven’t heard of before, which is thinking of your IT as a team member. Just as in any relationship, [where] you have to learn and you have to invest time in figuring how to work well with any of your colleagues, you need to invest time and energy and behave differently to work well with your IT system.

You’re also making the case that investing energy in being a better team member in general is an important thing for physicians to do for the era ahead. Am I interpreting you correctly?

Shah:  Absolutely. We know that those practices that perform well for their patients and capture a number of behaviors better than others that perform on average not as well [include things like] creating deeper patient relationships with their patients, leveraging decision support, closing care gaps, using team huddles every day, practicing informed shared decision-making, and treating patient complaints like gold.

These are things that are hard to do if you’re a solo practitioner, but if you have a team and different members of the team can take on different parts of that overall goal and overall work toward getting patients toward their goals, that can help us advance the care we provide exponentially to each individual in our practice, without extending the hours we stay open, the amount of work, or the cognitive burden on any individual.

The example I’ll give you is segmentation. We know that we should treat different people differently, and yet we start with a one-size-fits-all purpose for many of our practices in health care. When we close care gaps, we usually look toward the U.S. Preventive Services Task Force Grade A and B Recommendations to say “you need a flu shot,” “you need a mammogram,” and it’s a coarse way to understand what a patient needs relative to their overall total health. But it’s the best we have today.

Imagine a future, in the very near future, where we could segment patients not only on their clinical needs, but also on their capabilities, on their contexts, on their environments, on the social determinants of health they are facing. Folks who are starting to do that are seeing exponential gains and improvements in outcomes.

When you address in a dual Medicare-Medicaid population the social determinants of health like food insecurity, transportation, financial issues, and housing insecurity, those nonmedical issues lead to improvements in the medical outcomes of our patients — but it’s hard to do in the current environment. I hope that with better segmentation, with better aligning of resources to the patient’s needs, we can do that at scale in a way that we know holds much promise for America.

Lee:  You make a good case that working with AI and our colleagues, physicians can help deliver better care to patients, close gaps in quality, do a better job on many levels. But what’s special that we physicians can do, should do, must do? Because everyone loves hearing flattery, I want you to lay out what you think is remarkable about us clinicians that only we can bring to the party. One of the things I’ve liked in hearing you talk and in reading your words on this topic is that you have a strong sense of what physicians should double-down on as they think about the future.

Shah:  We’re seeing a lot of industries being changed fundamentally overnight with the introduction of AI and different business models. What our superpowers are as physicians include critical thinking, communication, collaboration, and creativity. These four Cs are going to be the most important skills to continue to develop in the age of AI and technology.

What we can do in hearing our patients, in caring for them, in helping them get from where they are to where they need to be and working with others to help them get there, that is uniquely human, at least for now. I’m hoping that for the next few years, if physicians continue to do what they’re good at and spend time on their superpowers —  critical thinking, communication, collaboration, and creativity — that will allow us to live in harmony with AI and see the benefits of these new technologies for both ourselves, our own workflows, our own lives, as well as for our patients.

Lee:  That is a wonderful perspective, which is going to be refreshing for a lot of clinicians. Maybe for this last part you can give an example of each of those four Cs so they’re vivid in the minds of our listeners and so they can work harder on strengthening what they’re already strong at.

Shah:  As we embrace our Millennials who are finally starting to get care, we know that they value different things than some of the older generations in how they get care and what they want from the care they receive in the health care setting. I’ve found that Millennials value convenience and access over things like privacy, and that when we have transparency in fee schedules and other things, those are things that we took for granted in my generation, but Millennials want to know up front before they even enter the office room.

So how do we get there? How do we get Millennials what they need? That takes a lot of changes from our current practice of medicine to the practice they need us to be. That creativity requires us to think about what makes sense from a retail model in the health care setting. What are the ways that we communicate to Millennials? If we don’t have email or text messaging or even Slack or other systems, if we don’t have access on evenings and weekends, Millennials are not going to find our practices as helpful as they want them to be.

Critical thinking means understanding the patient in the full context of their lives. They may say one thing, their spouse may say another, but you understand what their needs are given their prior choices. That critical thinking around what is going to lead to the outcome the patient needs is something that we have to do beyond reading between the lines of what the patient is saying.

Communication involves more than just speaking back and forth. It’s all the nonverbal. It’s the human touch. It’s the ability to understand the context that a patient lives in and then make that real in terms of how they are going to change their lives to achieve the outcomes they want.

Collaboration across our broader patient teams including the patient, the family caregivers, our own staff, and the systems they live and work in, that collaboration toward outcomes is something that again takes a unique skill set that only physicians have.

And finally, creativity. We are changing the context of how we care for people every day, and every day requires new solutions to new problems. We don’t have reams of data to go through to look at a similar situation on every patient and understand what the best outcome is; we have to make it up on the fly. Creativity is one of the superpowers that most physicians have and is what brings us to this field in the first place.

Lee:  That perspective that, yes, physicians need to adapt, they need to be ready to work with AI and [think] of AI as being like part of teamwork, that is a brilliant perspective. And your advice to feel good about the four superpowers that we have and to strengthen them, that’s great advice for the future.

I know that you’re going to be at that interface of innovation and health care delivery for the rest of your career, and I’m hoping we’ll be able to tap into your [thoughts] again, but your practical advice for facing this overwhelming issue for a lot of clinicians is welcome, so thanks very much.

Shah:  Thank you, Tom.

Call for submissions:

Now inviting expert articles, longform articles, and case studies for peer review


A weekly email newsletter featuring the latest actionable ideas and practical innovations from NEJM Catalyst.

Learn More »

More From Care Redesign
Community Resource Referral Type

Assessing and Addressing Social Needs in Primary Care

Lincoln Community Health Center improved care quality by measuring and responding to upstream social and economic risk factors disproportionately affecting low-income households.

Time Driven Activity Based Costing for ECMO

Achieving Value in Highly Complex Acute Care: Lessons from the Delivery of Extra Corporeal Life Support

To improve both the value and outcomes of ECLS, Cedars-Sinai Medical Center created guidelines for ECLS delivery and explored opportunities for more efficient care.

ARISE and SFHN BHVS Collaboration

The “Behavioral Health Vital Signs” Initiative

A safety net system’s trauma-informed approach to integrating interpersonal violence into behavioral health programs in primary care.

OpenNotes Epic Patient Email Cascade Chart February 1, 2018 - August 14, 2018

Measuring Performance of OpenNotes Initiatives to Target Improvement Efforts

How a New York safety-net health system used data science to identify obstacles to OpenNotes use, address technical barriers, and develop strategies for improving clinical note sharing by providers and viewing by patients.

Mapping a Technology Strategy for Bundled Payment Care Using a Value-Driven Framework

Harnessing Emerging Information Technology for Bundled Payment Care Using a Value-Driven Framework

A four-part framework developed by physicians at Partners HealthCare provides a stepwise process for assessing and integrating technologies to effectively use data through a continuous patient experience.

UCLA Health CKD Risk Stratification and Management

Proactively Catching the Declining Patient

A coordinated effort by UCLA leaders to identify a high-cost population with chronic kidney disease and to modify care processes and personnel has led to improved health and reduced utilization.

Telehealth and remote monitoring are little used and ineffective for chronic disease care

Survey Snapshot: Treating Chronic Disease Proactively

Though survey respondents don’t indicate strong use of telehealth and remote monitoring, NEJM Catalyst Insights Council members discuss the ways they’re using these tools to monitor chronic disease, with good results.

Platforming Health Care Operations - Consumer-Driven Health Care - Business-Minded Optimizations

Platforming Health Care to Transform Care Delivery

Health care leaders need to focus less on ownership and control of the delivery process, and more on outcomes, cost efficiency, and customer experience.

Shah05_ integrated systems innovation pullquote

Build vs. Buy: What Should Health Systems Do?

The consolidation craze continues, but vertical integration has yet to demonstrate real progress toward the Triple Aim. Health care leaders would do well to consider innovative approaches that are working in other industries, including the tech-enabled full stack model.

Diagram Illustrating the COPD Care Pathway at Allegheny General Hospital

End-to-End Care for COPD Patients that Improves Outcomes and Lowers Costs

Allegheny General Hospital created a comprehensive solution for patients with chronic obstructive pulmonary disease (COPD) that led to improved clinical outcomes, reduced hospital admissions and readmissions, and a resultant decrease in the total cost of care.


A weekly email newsletter featuring the latest actionable ideas and practical innovations from NEJM Catalyst.

Learn More »


Design Thinking

20 Articles

Taxonomy of the Patient Voice

While health care pursues the important trend of putting patients at the center of care,…

Measuring Performance of OpenNotes Initiatives to…

How a New York safety-net health system used data science to identify obstacles to OpenNotes…

Proactively Catching the Declining Patient

A coordinated effort by UCLA leaders to identify a high-cost population with chronic kidney disease…

Insights Council

Have a voice. Join other health care leaders effecting change, shaping tomorrow.

Apply Now