At our recent Tech Summit, I spoke about one of technology’s top trends, artificial intelligence. It’s playing a bigger role than ever in health care. You just might not recognize how.
From improving health care processes to predicting when you might need to go into the hospital, AI is improving many aspects of the way we obtain and pay for medical care. Most patients aren’t aware – yet – of what goes on to make AI a reality in health care.
But it won’t be long before health care providers and insurers unlock even more AI capabilities to help improve the quality and cost of care. Here are my thoughts on how this will happen.
A Brief History of AI
For many Americans, the concept of AI was introduced in 1968 with the release of 2001: A Space Odyssey one of the most influential sci-fi movies of all time. The story predicted that by the dawn of the 21st century, we would co-exist with intelligent machines that could think like us. And as the movie told it, those machines might even try to out-think us.
Well, 2001 came and went without the emergence of human-like robots. Yet, if we look beyond the inflated expectations, AI maturity and growth has gained plenty of steam in recent years. And for people in tech careers, that’s exciting.
Let’s consider why AI growth has accelerated and the implications for health care.
In large measure, it boils down to money. Large organizations have invested heavily because they see business potential in AI. For example, IBM has invested $15 billion in Watson and related products. Google has invested $3.9 billion in AI acquisitions alone.
Startups are also a good indicator of technology growth. Venture capital investment in AI moved slowly and steadily in the ‘90s and 2000s. Then, in 2012, investment in AI exploded.
AI growth also has been fueled by the maturity and availability of AI software, especially open source platforms and libraries. With literally $0 software investment, any organization can install, learn, and deploy highly-capable AI solutions. Of course, hardware grows faster each year and many companies, such as Intel, are building computer processors that are optimized for AI. More powerful hardware, highly capable software, and extensive financial investment have created the perfect conditions for AI.
Three Areas of Impact in Health Care
We’re seeing the dividends from all of that investment today. That’s especially true in health care, where AI holds enormous potential to help doctors provide more effective care, to lower costs and to make the system easier for consumers to navigate.
Let’s consider a few specific areas where AI is having an impact on health care: improving and managing health, process automation, and clinical diagnosis. We’ll mainly focus on machine learning, which is a type of AI that learns from data without explicitly being told how to do so.
Managing and improving health
The deployment of electronic medical record systems, fitness trackers and patient registries, to name a few, have led to much richer patient datasets. Such datasets, when used appropriately and consistent with privacy laws, can feed machine learning algorithms to improve patient care and health.
For example, Blue Cross NC recently piloted a machine learning model that was able to predict, with a degree of certainty, when a patient is likely to experience a health event requiring readmission to a hospital. This example demonstrates how technology can assist health care professionals in caring for patients and improving their health.
Reducing costs and improving patient experience through process automation
According to the Commonwealth Fund, about 25 percent of hospital charges are consumed by administrative functions such as billing and claims processing. When you add administrative costs incurred by insurers – which average about 10 percent to 15 percent of revenues – you can see health care is ripe for the improvements in efficiency that AI can bring.
Improved data sharing between health care providers and payers, along with machine learning methods, hold the promise of reducing manual checking and administrative work. Today, for example, providers must contact insurance companies to gain approval in advance for certain medical procedures. Imagine a future where data sharing along with machine learning greatly reduces the need for manual intervention for prior approvals—the system can learn over time to select the small set of cases that most require intervention.
Diagnosis of health conditions
Consider this example: A study published in 2017 demonstrated that, by training a system with 130,000 images of skin lesions, the system was able to detect skin cancer at a level equal to board-certified dermatologists.
Consider also an example from Japan. Two years ago, a woman was suffering from leukemia. The doctors were frustrated that their treatments weren’t working. Enter IBM’s AI platform, Watson.
Watson looked at the woman’s genetic information and compared it to 20 million clinical oncology studies. Using what it knew about the woman’s symptoms and the studies, the “robot” identified the specific type of cancer, an incredibly rare form of leukemia.
The doctors had been treating a different kind of leukemia, which is why the treatments weren’t effective. Sure, a team of qualified doctors could review and analyze 20 million studies – but that would take decades to do what a machine can do in hours. Watson didn’t trump human judgment, it simply did a lot of the legwork that enabled the doctors to design the right treatment plan.
The humans used Watson’s intelligence to do what they do best – save lives. This woman survived because a robot identified her cancer correctly and medical professionals acted on that intelligence to provide the right care.
Impacting People’s Lives
That’s exciting to those of us who spend our days applying data and technology to improving health care. It’s a privilege to work in an industry and for a company that sees such high potential in AI. Where else but health care can you see technology so directly impacting people’s lives in such positive ways?
AI and robots won’t replace people. We need people to train the machines – without fear that 2001: A Space Odyssey will come true and the robots will outthink us. We still need humans to use their brains to take the insights that technology gives us and convert those insights into effective patient care.