The Automation of Healthcare: 10 Questions We Should Ask about the Use of Clinical Analytics

Predictive modeling and clinical analytics will eventually rule the healthcare roost—and we must ensure that a framework for safe and ethical use is created as it inevitably does.

As I wrote recently, digital health has a growing conundrum: the ethics of everything. This is one of my first articles on the topic, originally published September 15, 2014.

The future of Big Data in healthcare is both bright and scary. And the use of predictive modeling in the clinical setting—aka clinical analytics—may be one of the most exciting and threatening issues we have to deal with. The July 2014 issue of Health Affairs, “Using Big Data to Transform Care,” provides an extensive and in-depth look at a variety of issues that healthcare will have to deal with as we surge forward into Big Data’s realm.

As the authors of “The legal and ethical concerns that arise from using complex predictive analytics in health care” put it:

“Predictive analytics promises to make dramatic changes in the way health care is practiced and delivered. The question is not if the health care system must be prepared to address the legal, policy, and ethical challenges that predictive analytics raises, but when it will be ready to address them.”

Honestly, by the time I was done reading through the entire issue, I was overwhelmed and more than a little unnerved.

Basically, there are more questions than answers at this point—and if you haven’t gotten your hands on the July 2014 issue of Health Affairs, you should. There are many questions to ask, and here are just a few. Some are iterations of what the various authors discuss, and some are what I see as both exciting and troubling possibilities.

1. Who (or what) is behind the wheel? 

The driver for the use of predictive analytics unfortunately may not be about optimizing the quality of care for the sake of quality itself—but about the bottom line of business and the allocation of resources.

In the aftermath of disaster, a prime responsibility of healthcare workers is the task of triage—determining the status of survivors and allocating limited resources to those most likely to benefit.

“Allocation of resources” will become the catch-phrase of the day in clinical analytics, and it’s unlikely clinicians will have much say about how that impacts the care of their patients. Hopefully, we won’t be in disaster mode with the use of clinical analytics—relying too heavily on algorithms to determine the definition of quality, and the means of determining who gets what.

2. What role will reimbursement play?

I see great potential for the insurance industry to jump into the clinical analytics fray as it becomes a more accepted model for determining care pathways.

Want us to pay for that extra day in the ICU? Show us your algorithm and we’ll show you ours—and tell you why we’re denying it.

It’s too early to know what that reality will look like, but we can be sure there’s some iteration of it in the future. Predictive analytics may seem like a boon for the efficiency of healthcare, but here, I predict it’ll lead to more payment-related hoops to jump through—as “predictive model decision support” becomes the phrase of the day.  

3. What choice do we have?

We have no choice but to plunge headlong into the use of Big Data and clinical analytics in healthcare. That train has left the station and the caboose is disappearing on the horizon. However, we do have a choice about how we do that in an ethical and safe manner.

Whether you see this glass as half-empty or half-full will depend upon how you interpret the fact that Big Data innovation is rapidly outpacing the development of a regulatory structure to ensure most appropriate use. 

Just as the need continues to grow for a new breed of healthcare experts who understand technology, so will the need grow for a new breed of privacy and legal experts who understand the role of Big Data in both.

4. Who’s the doctor?

Today’s healthcare system is changing so rapidly that veterans of medicine barely recognize its relation to the professions they signed up for.

To tell physicians that an algorithm will be directing the care they provide—and that they could be held liable for both refusing to follow accurate recommendations or following those that may be in error—may be too much to bear.

And to tell a patient’s daughter that the predictive model has declined an ICU bed for her mother who just had a stroke may be even worse.

The younger generation of medicine is undoubtedly thrilled at the potential that clinical analytics offers—but it leaves those of us grounded in the essentials of hands-on-care wondering if such a generation will know how to care for their patients when the system inevitably crashes.

5. Is “personalization” skewed?

Though it would seem that clinical analytics reinforces the shift to personalized care, it may in fact create the opposite.

This attempt at the personalization of medicine may actually be de-personalization of the very same—where algorithms built from a myriad of data points from a multitude of patients guide a patient’s care, instead of the patient’s own wishes in the context of the patient-provider relationship.

Sure, all of that data from all of those sources should be able to pinpoint specific need, but there’s nothing much more impersonal than telling a patient that it wasn’t the physician he’s had for 30 years who made the decision—but the algorithm in the Cloud above.

6. What perils will physicians face?

Many physicians already feel as if their practice is being dictated by outside influences—such as reimbursement structures and regulatory requirements.

But they ain’t seen ‘nothing yet.

Just wait until clinical analytics becomes the mainstream of care—and we’ll hope they still have enough clout to override an algorithm that their expertise and experience doesn’t support.

These early days may offer higher legal risks for physicians who do or don’t base clinical decisions on predictive models—and we’ll need a slew of data scientists who are well-enough-versed in medicine to write and interpret algorithms to ensure accuracy for care.

Yikes.

7. Will transparency be possible?

And what of the EHR? It’s the primary food source for clinical analytics, but what about the other end of the spectrum—where the algorithm spits out recommendations for care?

If you think interoperability has been a challenge among vendors due to proprietary issues, just wait until the wealth begins to flow from predictive modeling. Transparency about what factors influence the development of predictive models will be an ethical essential, but most companies are not going to be willing to share much more than the very basics.

It will be interesting to see how this dynamic plays out, and I’d be surprised if the U.S. healthcare system can manage to be a pioneer—since players here can’t even manage to move along the MU continuum in an effective manner. Leapfrogging environments—like the UAE and its GCC neighbors—may have the highest potential for making the most progress.

8. What role will healthcare institutions play?

And what about healthcare institutions?

Will they be required to reveal to patients the types of predictive models they use for allocation of resources?

Will patients be able to use this information in order to select the setting which best suits their needs?

In this age of increasing patient empowerment, the notion that we’ll be able to move patients along a passive treadmill of acceptance just won’t cut it—once the general public gains a greater understanding of how clinical analytics impacts just-about-everything across the threshold of acute and post-acute care.

9. Who gets what?

Depending upon how the implementation occurs, use of predictive models could widen health disparities if affordability becomes an issue—and with private sector development as the only option at this time, there’s significant potential for that.

That scenario foreshadows one in which government intervention will eventually be involved—which means more regulation—a consistent damper on innovation and progress.

10. How will consumers respond?

The coming cascade of consumer variations will most likely react according to generational context. The most resistant will be those who have one foot in the traditional approach to medicine and the other in medicine’s future—namely the Baby Boomers who will know just enough to sit-up-straight when the future invades the present and question such changes with a loud and collective voice.

They will help to shape the ethical use of clinical analytics going forward, as younger generations work behind the scenes to create the models which are impacting them. By the time the young ones are in need of more healthcare, they’ll be so embedded in the predictive modeling mentality that the only consideration will be how to do more-better-faster.

I’m just another HIT nerd.

But I’ve been a nurse for over 30 years. Though I love technology—I love patients and providers more. If the former doesn’t serve the latter, I’ll be the first to stomp my feet and demand that it does. I’m all for progress, but we need to make sure we don’t put the digital health cart before the horse when it comes to the safe and ethical provision of patient care.

They don’t know what’s coming.

It’s most likely today’s consumers have no idea what’s about to hit them on healthcare’s Big Data front. I doubt that many healthcare professionals do. The fact that there will someday be no differentiation between the terms “Big Data” and “healthcare” is the most significant prediction of all. Predictive modeling and clinical analytics will eventually rule the roost—and we must ensure that a framework for safe and ethical use is created as it inevitably does.

The nuviun blog is intended to contribute to discussion and stimulate debate on important issues in global digital health. The views are solely those of the author.