As AI, virtual care reshape the healthcare workforce, remember the difference between 'tasks and jobs'

Accenture's Kaveh Safavi discusses healthcare's outsized spend on labor – and how machine learning, telehealth and consumerism are transforming how work is done and who does it.
By Mike Miliard
10:45 AM
AI and virtual care are changing the healthcare workforce – but there's a difference between 'tasks and jobs'

As healthcare faces the realities of both a shrinking workforce and an expanding patient population, hospitals and health systems are faced with some fundamental challenges.

And they're spending lots and lots of money to help manage them. Accenture data shows that nearly two out of every three dollars spent by U.S. hospitals – 65 percent of expenditure, on average – goes to labor costs.

Like so much else in healthcare, that's set to change in the years ahead – is already changing – thanks to the twin revolutions of technology and consumerism.

Virtual care, telemedicine and remote monitoring are leading the change in how care is delivered and who can deliver it, ushering in the potential for huge savings in time, cost and efficiency while widening access to care for the patients who need it most.

Patients too are playing a bigger role in their own health, with more and more of them – from Millennials to the Silver Tsunami – showing an appetite and an aptitude for digital health and "self-service" care. (More than half of respondents to a recent Accenture survey said they've already made use of wearable devices so far this year.)

And, of course, artificial intelligence and machine learning, even at this early stage of their development, are setting the stage for major changes in the ways healthcare professionals do their work, whether they're frontline clinicians or back-office staff.

Dr. Kaveh Safavi, Accenture's global head of healthcare, spoke recently with Healthcare IT News about the big trends he sees on the horizon for the healthcare workforce – and the changes that already are taking place.

Q. What are some of the biggest changes you see in the next year, two years, five years for the healthcare workforce?

A. First of all, there is a shortage issue that is becoming more and more significant. The demand for services is growing faster than the labor supply is, particularly for skilled labor – anyone with a license. And you're starting to see certain kinds of areas where shortages occur. People worry now about nursing retirements, etc.

This is a global phenomenon. There isn't a country I work in where people aren't worried that the workforce just simply can't keep up with demand. You see it in poor countries and in rich countries.

So the first problem is that the supply of labor to provide services will not keep up with the demand for services if we keep doing it the way we do it today.

Another observation is that every other industry – and now healthcare – will be able to take advantage of artificial intelligence to start substituting for some of those tasks.

The good news is that in healthcare we're not worried about creating a reduction in the number of people working. We're looking at this as a way to expand capacity to close the gap. Healthcare, unlike others, as a bridge between the shortage of staff of labor necessary and the demand for labor.

Artificial intelligence is so important because much of what people in healthcare do falls into the concept of non-routine tasks, or not-completely-routine tasks. And artificial intelligence is really the first information technology that can begin to substitute for those kinds of non-routine tasks because it has the capacity to learn without explicitly being programmed by people.

With machine learning and deep learning, we now have an IT tool that can begin to take on some tasks over from the labor force in healthcare.

And that is both a cognitive and a physical aspect. By that I mean some tasks that nurses or pharmacists or even doctors might do, like information gathering or simple routine advice.

And then on the physical side, we see for example that where they can't fill nursing jobs, they have physical robots that help make rounds in the hospital to collect data and then make the data available for clinical judgment.

And so this idea that you can have both physical and cognitive tasks performed by machines as opposed to people is important.

And I would say tasks more than jobs. We don't really think about this as a job substitution. We could think of it as a task substitution.

And then the third one is really interesting, which is that there is an opportunity here to take individuals who might have had their jobs affected in other economies in other parts of the economy, and with a little bit of training and a smart device convert them into a frontline healthcare worker.

We've seen examples of this right now not in developed markets like the United States but in emerging markets in Southeast Asia for example, where you can give a lightly-trained worker a smart device and they can go become a frontline worker where they can check the patient.

In this case, they were using a device that would examine patients' wounds and bedsores, and be able to make some judgment as to whether this patient needs an actual expert to be seen, or whether or not this modestly trained caregiver can actually change the wound dressing.

You would never have been able to train up enough people to do that, but with a little bit of training and a smart device, suddenly they become frontline healthcare workers and you begin to close the gap that way as well.

Q. So to this idea of a difference between tasks and jobs: Is that something people can bank on? Because there are folks out there who are worried about jobs, whether it's a back office worker who's worried about being automated out of existence or a radiologist who sees machines reading images better than humans. At one of our conferences recently, someone said he could envision machine intelligent algorithms running an ICU.

A. The people who have looked at this and tried to measure a couple of places if you look back at that work that was done at Oxford in 2013, there was a big study that said 47% of U.S. jobs will be automated. People threw that quote around, but if you actually look at that paper they rate jobs based on probability of total replacement. And jobs that have clinical judgment are less than 1% probability of full replacement, compared to other jobs which have 99% probability for replacement.

We did our work looking at clinicians and clinical tasks, and our estimate was around 27% of the tasks that are done by clinicians can either be moved to patients or smart devices or a combination – but that's tasks, not a whole job. 

And also I saw some work that was done in the last two or three months by either RAND or Brookings, and for the clinical workforce in healthcare they put it at about 30% of tasks.

The only people who are likely to lose their jobs are people who, if you would look at their job today, you'd say, essentially, they are so simple in their nature that they're fully automatable.

But even if you look at, for example, healthcare management that insurance companies do, matching people with their benefits and directing them to disease management programs – our sense is that you're really just creating about 30% extra capacity (with AI) and those people get to stay on the job and do other work.

They don't actually lose their jobs because there is so much more work for them to do. Their work becomes different work. So I would say for healthcare in particular, the likelihood of people losing their job is extremely low – but for many people the likelihood that their job changes could be modest to significant, depending on what percentage of their tasks can get moved to machines.

Q. As John Halamka has said: If your doctor could be replaced by AI, your doctor probably should be replaced by AI.

A. That's maybe a crude way of putting it, but yes, I think very few doctors are doing work that can be completely replaced by AI.

If you look at, for example, what's happening within imaging –  it's not that the X-ray is being completely read, it's the fact that the machine is helping identify all the hotspots. That's giving the radiologist the ability to read more images. So you don't need as many radiologists, but what the radiologists are doing is they're moving toward more interventional and other procedural based things that don't need to be done with hands.

That's what's happened. They've redefined their tasks. They've redefined the nature of their practice. And frankly that's gone on forever. Cardiologists had to go through that – when heart disease and heart attacks became less frequent, cardiologists reoriented themselves to arrhythmias, for example.

Q. What should hospitals and health systems then be doing to prepare themselves for this future – whether it's with regard to where they're prioritizing their technology investments or whether it's preparing their workforce – the clinicians, the operational staff – to get ready?

A. First thing is to recognize that most of the first gains are going to come in non-clinical functions, rather than clinical ones. I think we spend too much time thinking about the clinical stuff because that's what makes good news. But the real opportunities for impact are going to occur in non-clinical areas.

For example, on the provider side all the things related to payment and reimbursement have a lot of people associated with them that are doing tasks, some of which can be shifted. Same thing with helping people find appointments right now. There's a lot of people involved in helping helping appointment-making, much of which is very crude and not very personalized. And we see technology helping to improve the productivity of that process, as well as the effectiveness of that process.

The next part is to recognize that this is not just about bringing IT into the organization. You actually have to change the process to get the benefits of it. We have watched other industries try to automate just using basic automation automating an existing process. They never see the gains that they were hoping for. They have to think about changing the process and the technology at the same time.

The third piece is when a human being works with a machine or artificial intelligence as a co-worker, there's a whole set of skills that have to get created, that have to be acquired by the human part of that human-machine dyad that we don't have today.

For example, you actually have to learn how to ask your smart technology the right questions to get the answers back you want. It's not just naturally intuitive. In the manufacturing world where you've already had humans and machines working side by side together, they've actually coined the term "cobot" – that's a robot coworker. And they've had to start figuring out how do you train people to work side by side with a robot.

The same thing is going to happen in the more cognitive areas. But there's a whole set of skills – we call them middle skills – where humans and machines work together that all have to be developed.

Q. Talk about telehealth, remote monitoring, virtual care – which all seem to be gaining steam after years of effort. How will they impact the healthcare workforce going forward?

A. We used to think about telehealth mostly as about solving for distance: Doctor in one location, patient in another. I use the term "virtual" rather than "tele" because I'm trying to send out a message that this is really about using a physical and digital combined platform for care – where you can get benefits that are more than just distance.

You can get benefits like one-to-many services, for example: one doctor and many patients  – group visits. You can also break the link of time, with asynchronous care. You can also augment the visit, which creates a different kind of outcome with information layered on top of it.

When we start thinking about virtual health as not just solving the needs of the unserved but the already-served, and create augmentation, we're going to see virtual health have an effect on productivity and affect outcomes.

Most of the time people have thought about telehealth as primarily making better, not about making care more affordable and cost effective. A few years ago we did some work on this and that became the genesis of our theory that about 30 percent of physician capacity can be moved to either patients or machines.

We looked at how much of a primary care visit can actually be done by offloading into something to either a patient or a machine or a patient with a machine. That, to me, is the way to think about virtual healthcare – the totality of it, rather than just traditional telemedicine thinking.

The concept of monitoring and care in the home is important too, because we see in care moving more and more to be location independent or location agnostic. The technology makes it possible to do things at home.

Look at the hospital-at-home movement. That's a national goal for us, as well as in the Netherlands which is really what I would consider it one of the global leaders. They're rebuilding hospitals now, explicitly with 40 percent less capacity for the same population, with the idea that they're going to use technology and human labor in a different way to be able to keep some percentage of those people at home.

To do that, you have to actually provide more care and monitoring at home and more coverage at home. But you can actually do it without bringing them into the hospital, and I think that's happening in the U.S. right now, primarily in the context of Medicare Advantage.

There's pilots going on in Boston around hospital at home where they can take a meaningful percentage of patients hospitalized today. And care for them at home much more intensely. You have to send equipment to the home where you can still take care of them less expensively than the hospital, probably more safely than in the hospital and decompress the hospital itself.

Q. And the good news, and we write these stories all the time, is that there's an appetite and a willingness to use these technologies – whether it's younger people with their wellness apps or seniors with at-home monitoring.

A. I was never worried about the wellness side of the question, I'm more talking about the acute care side, but the wellness side is even more expansive.

Q. So this future is exciting, but it's also clearly something that has to happen. You guys have pointed out that hospitals are devoting 65% of their spending on labor right now. What's not getting done because all that money is going to labor costs when it could be freed up for other more innovation?

A. That's a great way to think about it. There's two ways to look at it. One of them is that money being redeployed to make care better, but the other is just on a pure affordability basis. The data suggests that healthcare continues to lose productivity because it's so labor based.

And one thing that actually people don't recognize is if you look at the rate of increase of healthcare costs – not the absolute level of costs, but the rate of increase in the United States – it sits right in the middle of the OECD average and it's typical of European countries that have completely different insurance schemes and different pricing.

So we spend a lot of time thinking that our healthcare affordability problem is largely about bending the cost curve, that it's going to be solved for through how we insure and pay for care. That's actually not true. Because countries that do it completely different than we do are experiencing the exact same inflationary problem: Their costs are growing 1% to 3% faster than their GDP. In every country.

And that's because it's the labor model that of healthcare, which is true in all of these countries, that will grow at the same rate as GDP. And then when you add scientific innovation and an aging population, you're at GDP-plus. And nobody is going to stop innovation or an aging population.

So if we don't figure out how to do the same amount of work with less people, we're actually never going bend the curve. And that's not the same as taking waste out of the system. If I could take all the waste out of the system tomorrow and reset the baseline that would lower the cost on a per person basis.

But the growth rate would still be faster than GDP, because we're still doing it the same way. We're providing all the necessary care, primarily through labor and that's really the argument here. We have to rethink how we provide necessary care, not just simply focus on eliminating wasteful care, or over the next 20 years we won't change the cost curve.

Focus on Workforce Development

In July, we'll speak to experts about how they're managing their workforces – not just clinicians, but IT departments, data scientists, care managers, back office staff and others.

Twitter: @MikeMiliardHITN
Email the writer: mike.miliard@himssmedia.com

Healthcare IT News is a publication of HIMSS Media.