The Mayo Clinic has started a lot of artificial intelligence work, from its in progress alliance Use Google Cloud Joint venture to develop ECG-based algorithms. Now, the hospital system is establishing a two-year partnership with Verily to build decision support tools, starting with cardiology and diabetes management.
Although Verily and Google may work under the same parent company, the person in charge of the project emphasized that this is a separate job. Mayo Clinic cardiologist Dr. Rick Nishimura said the goal is to build a set of tools that are better than the options that doctors currently have.
Most tools today are “one size fits all,” and doctors need to figure out how to make software recommendations relevant to the patients sitting in front of them. They also did not solve the problem of displaying all the information needed for decision-making.Since there is little regulatory oversight for most clinical decision support systems, it is difficult to know whether they are truly accurate, or Too many false alarms.
“The current clinical decision support system, doctors hate… the computer says you must do this, otherwise we will continue to disturb you with these red lights in your EHR,” Nishimura said in an interview with MedCity News .
So, how do you structure what the doctor really wants to use?
For example, Mayo Clinic and Verily plan to design it to display patient records related to the case and obtain advice on medications, dosages, or tests from Mayo Clinic specialists.
“Doctors or nurses do not need to be told what to do, but to be able to accept these inputs and make their own decisions, which is more palatable,” he said.
Jane Shellum, head of Mayo Clinic’s IT department, said that first, these companies will work with subject matter experts to determine what is the most important information to help them make decisions. Then, they must find out where in the patient record that information is actually stored—is it in the echocardiogram report or in the clinical record?
After testing to ensure that they have the expected data, the next step is to present it to the clinician to see if there are other factors that are critical to their decision.
“I just talked to a cardiologist today, and he said that if you want to show me this piece of data and this piece of data…there is a whole story of data that tells what I’m thinking,” she said. “There is technical verification to ensure that the data is correct, and clinical verification to say, is it meaningful data?”
Later, they plan to conduct a study to determine whether the tool has successfully reduced the cognitive burden on clinicians.
Although they may start with cardiometabolic diseases, Shellum says they are designing the tool so that it can be used in different environments. Verily’s head of health informatics, Paul Varghese, said this is a good starting point because of its potential impact while still being “appropriately difficult.”
“It needs to synthesize a series of information that may be captured in different formats and different forms in the EHR. It may be encoded in structured data. It may be fully recorded in the clinical record,” he said. “When we looked for a solution for this, we realized that it requires a reasonable level of complexity, knowledge and experience.”
They plan to design it in such a way that it can be integrated with most commercial electronic health records using Smart on the FHIR standard. First, it will be deployed inside Mayo Clinic, but the two companies plan to expand it to Verily’s medical system partners and customers.
Photo credit: ipopba, Getty Images
