12 Ways Artificial Intelligence Will Transform Health Care
While the future of AI looks bright, there are several questions C-suite leaders must ask themselves.

September 28, 2015

David Ollier Weber


You may be familiar with the name Robert Wachter, M.D. He's written six books and hundreds of journal articles; he chairs the department of medicine at the University of California, San Francisco; and he's a leading advocate for patient safety. One health care magazine this year anointed him the nation's most influential physician-executive. He's perhaps best known for having coined the term "hospitalist," and for having defined and promoted hospital medicine as a recognized primary care subspecialty.
In a new book, The Digital Doctor: Hope, Hype and Harm at the Dawn of Medicine's Computer Age (McGraw-Hill Education, 2015), Wachter takes a deep dive into the turbulent waters of medical technology as informed by artificial intelligence, or AI. "While computers are preventing many medical errors," he observes, "they are also causing new kinds of mistakes, some of them whoppers."
It was one of those — at his own institution, UCSF — that inspired Wachter to examine where AI has got us and where it's taking us. Tiny oversights by a topflight pediatrician, a veteran pharmacist, a vigilant nurse and a poorly designed computer order entry system combined to deliver a massive overdose of antibiotic to a 16-year-old patient. The harm was reversed, but the lesson was clear. AI in health care still has a long way to go.

Looking over the Mountain

Artificial intelligence is what enables a digital device to see and recognize objects (e.g., read a bar code), understand and reply to normal speech (à la "OK, Google"), make decisions and even learn to change its thinking and behavior as it analyzes the gazillions of data bits in the distributed memory known as the cloud (viz. IBM's Watson). AI infuses the modern health care system.

And yet, said Wachter in a recent telephone interview, "I'm incredibly disappointed in where we are. I'm a big believer in technology. I tweet. I blog. I'm spoiled by technology. It's supposed to make things work better!"

And it will, he believes. Eventually. "I'm fairly optimistic," he acknowledged. "There is a happy ending."

However, as what's termed a disruptive innovation, AI will certainly inflict casualties along the way. This is what Wachter sees when he "looks over the mountain at how this might end up working":

Many of the nation's smaller hospitals will close, avers Wachter, because they will be left behind in the race for quality by competitors quicker to adapt to cognitive computing and cloud-based AI technology. In the wired environment, geography won't matter much.

Same for most small independent physician practices.

If you feel sick enough to need urgent care, you will go to a conveniently located clinic, probably in a local mall or chain pharmacy. There you'll be seen by a nurse practitioner working "at the top of her license" and able to take into account your entire medical history by pulling up your universally accessible, privacy protected, electronic health record, or EHR.

You won't have to leave home to get advice about how to treat many worrisome conditions. Simply dial your health care provider on a smartphone and send a picture or a video of, say, your child's inflamed ear. A computer will read the image and recommend how to proceed. (Thanks to machine learning, the computer's better at pattern recognition than the human eye/brain. The same goes for decoding X-rays, skin rashes and biopsy slides. Wachter predicts that radiology, dermatology and pathology are three of the medical fields most likely to be swept under by "the digital tsunami.")

Similarly, most people with chronic conditions will be cared for at home by nurse educators and doctors who pop in frequently — just not in the flesh. They'll chat with you via the television set, laptop, tablet computer or smartphone, acting on data from implantable, wearable or external sensors. Shut-ins may be watched over by furry AI-imbued robots that double as caregiver/companions. (See Part 1 of this series.)

If you do need to be in a hospital, you will be very, very sick — or there for major surgery or for diagnosis of some puzzling, rare or extremely complex condition. Most people who are infirmed will be cared for in less expensive, more comfortable settings.

Thriving hospitals will be huge and "bristling with technology." There will be no such thing as an ICU; every single room in the facility (and they will all be single rooms) will be a self-contained ICU.

Nurse staffing ratios will be adjusted constantly according to the individual patient's need as determined by AI risk-monitoring and treatment algorithms. Each room will feature a large video screen on the wall for display of the patient's EHR and for interactive tutorials using computer voice recognition and prompts. A variety of built-in cameras will enable close-up examinations of the patient or wide-angle views of those in the room during remote specialist consults. Clinicians won't have to be physically present (but they'll have to be licensed nationally or even internationally).

Most physician orders and notes will be entered into the EHR through natural language voice recognition software while the kindly doctor looks you in the eye. Each patient will control his or her own EHR, a digital compendium of "clinician-generated notes and data with patient-generated information and preferences." Redundant information requirements will be eliminated.

Alerts will be calibrated to clearly distinguish life-threatening problems from minor anomalies — a blurring that plagues clinicians in today's hospital environment and a major contributor to the errors that inspired Wachter's book.

Physicians will be aided in differential diagnosis and evidence-based treatment by cognitive computing systems like IBM's Watson. Artificial intelligence applied to cloud-dwelling Big Data will assist clinicians by juxtaposing the individual patient's characteristics — down to the last nucleotide in her genome — against billions of anonymous, equally detailed patient histories and the latest findings in world medical research (which will no longer rely on expensive and elaborate clinical trials.) Easy-to-read dashboards will track adherence to each step of the tailor-made care plan.

A new health professional will emerge "akin to an air traffic controller." Not a physician but under the supervision of a physician, his or her task will be to "understand the data, put it in context and act on it."