By Jon Elwell
For many of us in health care, artificial intelligence (AI), or machine learning, as it’s also known, is the bright, shiny object we have our eyes on right now.
There are good reasons to be excited about the technology, not the least of which is its potential to transform the vast amounts of big data we are accumulating into intelligent insights that weren’t possible even a few years ago. The promise of AI is that it can identify patterns and trends hidden within the data that can tell us what we should be looking at—and even what we should be doing about it.
There’s only one problem with this focus on all of the wonderful things that AI can do: It ignores the current realities of the state of health care data—even at some of the most forward-thinking health care organizations in the United States.
Facts About Fax
Taking advantage of AI begins with the assumption that all of the information organizations need already exists in discrete data and is being exchanged electronically, system to system. That may be true within acute care hospitals, primary care practices, and specialists’ offices that either voluntarily implemented EHR systems or were required to do so under the Affordable Care Act.
For postacute care facilities such as skilled nursing facilities and assisted living communities; physical, speech, and occupational therapists; diabetes educators; dialysis centers; emergency services; and other vital providers that make up the rest of the care continuum, that may not be the case. In fact, as anachronistic as it may seem, many still use paper faxes to exchange patient information. The result is that the futuristic breakthrough that we’re all relying on to drive personalized medicine is being limited by a technology from the 1980s: fax.
One of the biggest issues with fax is that organizations can easily send care documentation, whether via paper or electronic fax, but it is very difficult for them to receive and efficiently use that information. Although solutions exist that can “scrape” information from PDF or TIFF files, they are very unreliable in terms of accuracy. All it takes is mistaking a “5” for an “S” to make the incoming information invalid.
The result is that either a human being must read the fax and input the data manually, or the data don’t get entered into the EHR at all. Either option is undesirable if the goal is to make AI work.
Taking Steps Toward AI
So how do we make that leap, particularly in these difficult financial times when many providers that have an EHR still haven’t achieved sufficient interoperability? The movement of discrete data remains elusive. As in the Chinese proverb, this journey of a thousand miles begins by taking a few simple steps.
The first is for those providers to move away from paper fax and toward the growing Direct messaging network. This network enables providers to make a secure, system-to-system connection so patient data can go directly into an EHR rather than relying on humans to input the data.
Of course, the information has to go to the right place—or system—to be useful. One of the key benefits of Direct messaging is that there is already an established, secure electronic data exchange network incorporating 1.5 million Direct addresses/accounts across more than 94,000 health care organizations, along with a single directory that makes it easy to look up and confirm where the data should be sent within normal workflows. As with any network, the more members that join, the more valuable it becomes. Ideally, every provider in the world will eventually take part.
It doesn’t matter which technology the senders and receivers are using, either. Direct messaging allows the data to be securely exchanged so they can be easily consumed and added to the rest of the patient’s record—all without any human intervention.
Solve interoperability and the value of investing in AI becomes exponentially greater.
Walk Before You Run
There is no doubt that AI is poised to be a true paradigm shifter. But first things first: Before we all jump on the AI bandwagon, it’s important for health care to get its interoperability house in order to ensure that all of the data needed to power AI are ready, available, and consumable. Otherwise, it will be just another technology that fails to live up to its promise.
— Jon Elwell is CEO of Kno2 LLC.