도서요약

AI Transforms Healthcare

The game-changing technology of open-source artificial intelligence that we discussed this month in Trend #3 will dramatically transform countless industries. Let’s explore how AI will unleash the biggest revolution in healthcare since the discovery of penicillin.

No industry is more ripe for disruption than the U.S. healthcare industry, which faces such problems as:

- Drug development obstacles that require pharmaceutical companies to pay billions to develop a life-saving new treatment, and then wait a dozen years for FDA approval.

- Inefficient twentieth century practices in doctors’ offices and hospitals that lead to long waiting times for patients and mountains of paperwork for doctors.

- A critical shortage of doctors at the same time that the population is aging and putting even more pressure on the health care system; by 2050, one in five Americans will be over the age of 65.

- Costs that are outpacing inflation and spiraling out of control. According to government estimates, Americans will spend $4.8 trillion on healthcare in 2021, accounting for 20 percent of GDP. That’s nearly double the $2.6 trillion spent in 2010, and 64 times the $75 billion spent in 1970.1

- A system that encourages doctors to order excessive tests and procedures in order to avoid malpractice suits. According to estimates by PricewaterhouseCoopers, as much as $1.2 trillion a year of healthcare spending is wasted on treatments that patients don’t even need.2

The good news is that innovations based on artificial intelligence offer solutions to all of these problems.

For example, an AI system could recognize speech, analyze images, absorb all of the knowledge in all of the world’s medical journals, detect patterns from a patient’s symptoms, and then produce an accurate diagnosis.

Not surprisingly, leading tech firms are working on AI projects targeting healthcare. According to IBM, by 2020, the amount of medical data is expected to double every seventy-three days. Wired.com reported that a doctor would have to spend 160 hours a week to read the new medical research that is published.3

Clearly, there’s no way that any doctor or hospital could keep up with that deluge of data. But IBM’s Watson supercomputer needs only fifteen seconds to read 40 million documents. It can also look for patterns in the medical records of 1.5 million patients. And since IBM recently acquired Merge Healthcare, Watson can compare any photograph, X-ray, or MRI of a patient to Merge’s archive of thirty billion medical images.4

With that power at its disposal, Watson can often make more accurate diagnoses than doctors. In fact, in a series of experiments conducted with the healthcare firm Wellpoint, Watson correctly diagnosed lung cancer in 90 percent of cases. While that isn’t quite perfect, consider that the success rate for trained physicians is only 50 percent. That’s no better than a coin flip.

The application that IBM developed, called Watson Health AI, is now used by sixteen cancer institutes in the U.S. to diagnose and treat patients, according to a post on BigThink.com.5
And according to HealthcareDive.com, “Watson is also partnering with Johnson & Johnson to use AI to help joint replacement patients better manage their health. It is also partnering with Medtronic with the goal of being able to predict which diabetes patients need more help in keeping their blood sugar under control.”6

Google has developed its own AI offering for healthcare, called DeepMind Health.7 The system will notify doctors when patients need urgent treatments, advise them of any potential drug interactions, and anticipate the health problems that a patient will face before those problems occur.

One early DeepMind Health application is called Streams, a program that enables doctors to see the results of medical tests instantly, which is unprecedented. For example, a pilot project at the Royal Free Hospital gave physicians the ability to find out within seconds of a blood test whether a patient was at risk of acute kidney injury. This made it possible to treat the patients who needed the most urgent care immediately, instead of waiting for results to come back from the lab.8

While IBM’s Watson and Google’s DeepMind could help physicians make better decisions and save lives, another artificial intelligence system could save costs. Praxis Concept Processing is an AI-enabled software program that captures what a doctor says or writes, and then matches the patient’s condition to the most recent similar case the physician has encountered.9

Some patients will have symptoms that are identical to those of previous patients. In that situation, according to the Praxis website, “If the closest encounter is identical to your present one, you are done! Praxis [Electronic Medical Records] will generate the new encounter instantly in the same way as you did it before. It will also print or fax your prescriptions, generate instructions to your patients, instructions to your staff, admitting orders to the hospital, procedure reports, work or school excuses, and every other document you require, all effortless and instantaneous. And, it will optimally code the visit and create a super-bill. All this happens in the blink of an eye!

“If the encounter is similar but not identical, you simply modify the differences from the closest case using handwriting recognition, voice recognition, or keyboard. The beauty of Concept Processing, however, is that it memorizes all your changes. So, when your next encounter falls between two similar cases, your editing is cut in half, and then by a quarter for the next case, and then by an eighth, etc. In fact, the more you use the Concept Processor, the faster and smarter it becomes. In time, Praxis becomes you?a true mirror of your mind.”

Theoretically, then, physicians can provide quick solutions to routine cases, freeing up more time to spend with patients with unique problems that need more of the doctor’s attention.

Meanwhile, Apple, Dell, Hewlett-Packard, and Hitachi are also reportedly working on developing artificial intelligence solutions for the healthcare industry. By 2021, the use of AI in healthcare is projected to jump to $6 billion, from $600 million as recently as 2014, according to research by Frost & Sullivan.

That includes the increasing penetration of AI into the drug discovery process. Artificial intelligence can evaluate every possible combination of compounds simultaneously to find new treatments that would elude human researchers tediously experimenting with one combination at a time.

For example, pharmaceutical startup Berg Health recently announced that a cancer treatment will reach the market after seven years in development - half of the fourteen years that are typical for new drugs - due to the use of AI in testing.10

The company used AI to compare fourteen trillion data points within tissue samples from patients with cancers of the pancreas, bladder, and brain to samples from people without cancer. The system identified the differences between the biological profiles and chose patients who would get the best results from taking the drug.

According to Berg co-founder Niven Narain, a clinical oncologist, bringing a new drug to market typically costs $2.6 billion and takes twelve to fourteen years; using AI, both the time and cost are cut by at least 50 percent. As Narain explained, “There’s a lot of trial and error in the old model, so a lot of those costs are due to the failure of really expensive clinical trials. We’re able to be more predictive and effective... and that’s going to cut hundreds of millions of dollars off the cost.”

To further eliminate wasteful healthcare spending, a company called Hindsait is launching an AI platform that allows doctors and insurance companies to flag prescribed treatments that are redundant or not medically necessary.11

In the emerging field of precision medicine, AI unleashes the potential of technology to deliver tailored solutions to each patient based on his or her unique medical history, lifestyle, and genomic make-up, rather than just the patient’s disease. When dealing with such data-intensive decisions, AI is proving far superior to humans.

How should companies respond to this opportunity? According to Anil Jain, M.D., vice president of IBM’s Watson Health division, leaders of healthcare firms should ask themselves the following questions:12

- “How can we aggregate all the data input generated within our four walls to do right for our patients?”

- “How can we draw insights from the data to take actionable steps to improve quality of care?”

- “How can we serve up the data to make each and every member of the care team operate at the highest level?”

- “How can we do all this cost-effectively?”

Jain concludes that companies that fail to embrace AI will end up on the losing side of the phenomenon that Austrian economist Joseph Schumpeter called “creative destruction.” As Jain puts it, “The incentives to buy high-functioning technology systems will be the same as for other businesses in competitive markets: Namely, the price for not doing so will be swift Schumpeterian death in the marketplace.”

Going forward, we foresee the following developments emerging from this crucial trend:

First, artificial intelligence will provide the solution to the looming shortage of physicians.

Within a few years, AI-based expert systems in the hands of masters-level “physician’s assistants” will be able to prescribe treatments for 90 percent of ailments with greater efficacy than today’s best doctors. This will free up the time of the limited supply of physicians to treat the remaining 10 percent of patients. According to Robert Wachter MD, chair of the department of medicine at the University of California, San Francisco, and author of The Digital Doctor: Hope, Hype and Harm at the Dawn of Medicine’s Computer Age, many patient visits currently handled by doctors will be routed to a new type of healthcare professional who will work under the supervision of a physician to “understand the data, put it in context, and act on it.”13 According to projections from the Bureau of Labor Statistics, the demand for so-called health information technicians will rise by 15 percent from 2014 to 2024, which is significantly greater than the overall estimated growth in the labor market over that period.

Second, many small hospitals and independent physician practices will go out of business when AI transforms healthcare.

They will fail to keep up with bigger medical practices that have the funds and the foresight to invest in AI technology, and the market will punish the laggards for their higher costs, less accurate diagnoses, and longer waiting times for everything from patient appointments to medical test results. The skilled professionals and capital assets of marginal players will be absorbed by the winners.

Third, AI will provide an affordable way to address the rising costs of elder care as the U.S. population rapidly ages.

A very promising option is to create AI-based living environments based on the Internet of Things along with targeted use of service robotics. AI could enable a small team of human caregivers to cost-effectively ensure the well-being of a large population of older people living independently. Humans would only intervene when the technology identified a problem.
Fourth, AI will dramatically accelerate the process of drug discovery by managing and tracking fully automated testing.

For instance, thousands of compounds can be tested on identical cells and the results correlated with the genomics of those cells. This will not only accelerate the discovery of optimized treatments for cancers, but lead to cures for other chronic maladies like Alzheimer’s, Parkinson’s, and heart disease.

References
1. For information about the rapidly increasing cost of healthcare, visit the Centers for Medicare & Medicaid Services website at:

https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/NationalHealthExpendData/index.html

2. For additional information about excessive healthcare costs, visit the Aetna website at:

http://www.aetna.com/health-reform-connection/aetnas-vision/facts-about-costs.html?TSPD_101_R0=7cb100b206709c2a01fad364fa0d4c2dw7O0000000000000000226ae5a2ffff00000000000000000000000000005758dc5500026457f5

http://www.wired.co.uk/article/ibm-watson-medical-doctor

http://www.forbes.com/sites/greatspeculations/2015/08/12/ibm-acquires-merge-health-to-supplement-watson-healthcare/-6f089d6a4f70

http://bigthink.com/philip-perry/how-artificial-intelligence-will-revolutionize-healthcare

http://www.healthcaredive.com/news/5-ways-artificial-intelligence-is-changing-the-face-of-healthcare/415424/

http://www.bloomberg.com/news/articles/2016-02-24/google-s-deepmind-forms-health-unit-to-build-medical-software

8. iBid.

http://www.extremetech.com/extreme/228830-the-next-major-advance-in-medicine-will-be-the-use-of-ai

http://www.telegraph.co.uk/technology/news/11920393/Cancer-drug-development-time-halved-thanks-to-artificial-intelligence.html

http://www.healthcaredive.com/news/5-ways-artificial-intelligence-is-changing-the-face-of-healthcare/415424/

http://www.hhnmag.com/articles/6561-ways-artificial-intelligence-will-transform-health-care