On a clinical level, American healthcare is arguably the most advanced in the world, powered by science and technology and guided by skilled and dedicated medical professionals. But the healthcare system itself is inefficient, byzantine, and often wildly expensive, due in large part to systemic inefficiencies that result in vast waste.
“The National Academy of Sciences/Institute of Medicine (IOM) issued a report a few years ago that said about 30% of our healthcare spending does nothing directly to help patients—and, in fact, can often harm them, which isn’t sustainable,” says Gregory Moore, Vice President of Healthcare for Google Cloud. “If we took half of that out of the system, we’d solve much of the healthcare spend issue for the U.S.”
Artificial intelligence (AI) is being hyped as the next big thing for, well, nearly everything, and healthcare is no exception. It has already shown promise as a clinical tool, and experts believe it is one of the keys to containing runaway costs in the healthcare system. Nick Giannasi, Ph.D., chief artificial intelligence officer for Change Healthcare, says that when computing tools such as machine learning are applied to many of the back-end administrative tasks that burden the healthcare system, there is an opportunity for significant savings across the entire system.
“Although it sounds far more sexy to talk about diagnostics and clinical care, there’s actually relatively low-hanging fruit with AI and machine learning to address one of the biggest problems in the U.S., which is the cost of healthcare,” Giannasi says. Coupled with new, potential ways of administering care at scale while also creating highly personalized patient experiences, AI is poised to dramatically alter the cost and also the culture of the entire healthcare system.
While waste and inefficiency are an unwelcome yet expected part of any large system, the scale of the issue in healthcare is astonishing, and increasingly crippling. “We see huge numbers reported,” says Giannasi. “$177 billion in fraud, waste and abuse; $91 billion in no-value-added work; and $35 billion in costs from poor communication between payers, providers and patients due to a lack of interoperability. And these are annual numbers,” he adds. The problems stem, in large part, from human limitations. Thus, a suitable solution comes by turning to machines to complement and augment human effort.
While humans continue to outshine computers in many realms, numerous aspects of the healthcare system are perfectly suited for the pattern-recognition prowess of computers. For instance, Giannasi says that a routine hospital-based medical procedure might generate a claim that requires human parsing of anywhere from a 400- to 1,000-page record of treatment for billing purposes, an arduous and time-consuming task. As a result, errors of over- and underpayment are so common as to be normal.
“A large payer has an army of people that basically do audits looking for overpayments. It takes a lot of effort, a lot of costly resources, and it’s slow,” says Giannasi. While painstaking for humans, crunching through such tasks is trivial for machines. “Instead of taking months, costing a lot, and having a poor accuracy or success rate, they can do it millions of times faster, thousands of times cheaper, at a higher quality, and at least more consistently,” says Giannasi. “It has the potential to free up a huge amount of dollars annually.”
Once these tools are adopted across the system, it’s likely patients will experience healthcare administration like they do other transactions, with instant billing and claims that lack all the clerical hassle and delays they suffer now.
Beyond administrative efficiencies, machines offer new mechanisms for engaging patients both at scale and on an individual level.
At the macro level, AI tools will enable highly granular preventative or maintenance outreach programs that in the past were functionally impossible at the same scale. According to David Gutelius, chief executive officer of Motiva AI, a company focused on AI for patient engagement: “There is an opportunity to become a kind of assistive, orchestration layer for the health care system at large. You have the machine that’s aware of resources in a large health care system and can essentially match those resources to the need, on the fly.”
In practice, that could mean providers might have automated programs that, in an instant, can identify high-risk patients for an emerging disease pattern within locales — or an entire population, for that matter—and deploy early intervention programs before the disease takes hold, creating better outcomes at a far less overall cost. These methods aren’t restricted to clinical roles. “The same sort of techniques could be applied to identify patients, or cohorts of patients, that are eligible for financial relief or support from other sources of funding such as eligibility for Medicare or Medicaid, or disability coverage,” Giannasi says.
Paradoxically, that same computing power could also be used to zero in for deeply personalized modes of patient interaction and engagement. This means deploying smart systems that track a patient’s condition, respond proactively to their needs, and then offer outreach early on instead of waiting for the patient to develop something serious.
“It’s about learning what a patient needs and prefers, and how to best serve her in a way that’s unexpectedly delightful. That might be through text messages or using an interactive voice response system, or a nurse. The experiences will be adapted and tailored to you as a patient,” says Gutelius. “That’s how we increase engagement in healthcare. And the more engaged patients are in their own health, the better the health outcomes and the more aware they are of the purchases they’re making in healthcare. By shaping the experience in a very personal way, we can actually drive value across the system, from patients all the way through to payers, while helping patients live happier, healthier lives.”
Artificial intelligence isn’t a panacea for the problems in the U.S. healthcare system, but the technology is sparking a rather large wave of transformation, and new and novel ways of applying data science to healthcare conundrums continue to emerge for every step of the patient healthcare journey. And as Moore notes, AI is just one tool of an overall holistic approach that will be required to create an efficient, cost-effective and patient-centric healthcare system.
“You can’t really solve the patient experience in a very detailed way until you go back to the basics and understand the root causes, some of which include siloed information and poor communication contributing to decisions being made very slowly, and enacted with high costs, both in terms of human suffering and dollars,” says Moore. “That’s where the cloud and AI can really start having an impact. These technologies can help providers deliver higher quality healthcare at lower costs and enable delightful experience for both patients and their caregivers.”
Change Healthcare is inspiring a better healthcare system. Learn more about our practical healthcare technology innovation strategies.
Developed in collaboration with CNBC Brand Studio.