The most effective healthcare technology must leverage the physicians’ existing skill sets. Artificial intelligence (AI) for healthcare providers is currently at the peak of inflated expectations, according to Gartner’s recently published “Hype Cycle for Healthcare Providers2019.” This is not a surprising conclusion, given the industry’s ongoing claims about disruptive AI-based solutions that will transform healthcare.

When you start reading about how “ambient AI” is going to see and hear everything that is going on in an exam room and somehow magically convert that to usable data for analytics and machine learning, you should be very skeptical. This misplaced dream of an all-knowing, all-seeing machine is holding developers back from addressing the real problem: giving clinicians tools they can use at the point of care today.

Before we defer to Dr. Alexa for all our healthcare needs, let’s first consider what type of AI innovations have the potential to improve healthcare efficiencies — and which functions don’t address what’s really broken in healthcare.

AI In The Exam Room

Many years ago, IBM launched Watson Health in hopes of solving medicine’s biggest challenges. Watson was going to evaluate a patient’s chart and use that information to find similar clinical presentations in other cases, literature or clinical studies, and then provide recommendations.

Despite spending billions of dollars in development and promotion, Watson failed to live up to the early hype. Watson was also unable to meet promises of providing superior diagnostic tools.

One problem was that Watson’s “brain” must be continuously fed the latest evidence-based content and protocols for every disease state, complex cancer, genetic test, medical therapy and more.

The brain of any machine is dependent on the data inputs and the algorithms used to process that data, combined with the logic designed to generate outputs. A primary weakness of any process is the precision and relevance of the data inputs, which is particularly challenging in clinical scenarios where time is of the essence and clinicians are already pressured to collect data to drive reimbursement rather than focusing on data to drive clinical care. And, replicating the brain of a well-trained and experienced physician is easier said than done.

More recently, several companies have announced plans for AI-based solutions that can be used in the exam room. These systems leverage voice recognition and/or video. In June, Saykara announced the first fully ambient AI healthcare assistant to see and hear what happens during the patient exam. The technology then generates a clinical note, including the patient care plan and relevant orders.

Later, if a physician wants to verify the specifics of a patient exam, they must wade through pages of transcription or watch the video of the visit. While clinicians might save time documenting encounters, they end up spending more time searching for critical patient information at the point of care.

What Physicians Do And Don’t Need

Physicians need technologies that streamline workflows and provide clean data that helps them assess the health of the patient in front of them.

They don’t need AI to help them with every aspect of the clinical encounter any more than drivers need GPS devices every time they get behind the wheel of a car. Having turn-by-turn directions can help a driver get to their destination sooner without getting lost. On the other hand, if a person is driving the exact same route day after day, perhaps from home to work, a GPS reminder to “turn in 800 feet” isn’t helpful. In fact, most drivers would be annoyed by such prompts.

In an article titled “The ‘inconvenient truth’ about AI in healthcare,” published in npj Digital Medicine, the authors say that “‘the inconvenient truth’ is that at present the algorithms that feature prominently in research literature are in fact not, for the most part, executable at the frontlines of clinical practice.”

Physicians need solutions to help them do their job better — not to tell them what they already know. Optimal healthcare technology leverages doctors’ strengths. An experienced physician rarely needs a computer algorithm to process multiple clinical findings to come up with a patient’s diagnosis or to decide to refer the patient to a specialist or another physician who may be more familiar with certain clinical presentations. A physician already knows how to do this.

Instead, physicians need real-time access to patient-specific information — for instance, tools to intelligently filter all the details within a patient’s chart and transform the data into actionable information for clinical insights. Take a complex patient with multiple chronic conditions as an example. The physician should be able to select any problem and quickly see the medications, lab results, therapies and other items that are relevant for that specific problem, such as hemoglobin A1c and metformin for diabetes, or serum creatinine and dialysis for chronic renal failure. By having the right data readily accessible, physicians can easily take advantage of the best computer in the room: the one between their ears.

As an industry, we need to focus on creating tools that fix what’s broken in healthcare — specifically, tools that streamline clinical documentation workflows. This will allow physicians to increase their productivity and have more time for direct patient care. For example, if a Medicare Advantage patient has a chronic condition that qualifies for increased reimbursement based on clinical risk, the point-of-care tools should identify those conditions and present the clinician with relevant information to monitor, assess, evaluate and treat the patient, while making it easy to document in real time and meet reporting requirements.

Physicians need solutions to take the tsunami of data within patient charts, format it as structured data for analytics systems, and create dashboards to support their clinical though processes.

More technology that fails to enhance the delivery of patient care is not worthwhile. Clinicians are among the most highly trained knowledge workers in any industry. They know what they need. Ask them. Listen. Then get to work on the solution.

See more at