Gruen said radiologists' jobs have become more complex in recent years, with far more images in data sets and multiplanar image reconstructions. Dictation is rapidly becoming a thing of the past as radiology moves toward voice recognition automated transcription directly into reports. And there are now additional AI notes or elevated reads on the DICOM lists. Workflows also are complicated by many distractions, such as the ED calling about stroke or pulmonary embolism cases, referring physicians wanting to see breast ultrasounds that go with mammograms, or other calls for quick consults throughout the day.
"Add to all this that we know radiologists error at a rate of 3%-5% on a national average, so all of this is ripe for technology workflows to allow me to do what I have to do, which is to keep my eyes singularly focused on the images and not miss that pulmonary nodule," Gruen said. "As a radiologist, my job should not be to answer the phone better, make sure my passwords are synched, make sure my hanging protocols work, or to have 17 different buttons to press to drag and drop something. We need technology to help radiologists to just stare at their pictures, think and take care of people. And the primary care doc should be able to look at their patients and listen and communicate, and not do what we have seen in so many cartoons where they turn their back on their patients to fill in the blanks in the EMR."
AI to aid radiology workflow
All vendors Radiology Business spoke with across the HIMSS expo floor said AI promises enable big leaps to improve radiology workflow.
"I see AI as an inevitable place where not just radiology, but all of medicine is going," Gruen said.
In radiology imaging, AI can help find small lung nodules, or help quickly identify and alert care teams about a emergency condition like stroke, brain bleeds or pulmonary embolism, and it can help serve as a second set of eyes to make sure nothing is missed. Or AI can look for incidental findings in areas that are not the reason for the exam. He said AI also can help manage complex reading lists to ensure the right radiologist is reading the right exams, while load-balancing to account for RVUs and prevent cherry picking. But he said AI goes so much further beyond this in nonclinical systems.
"Things like AI to improve billing, AI to get more people the healthcare resources they need, AI to make sure we are tracking the right people, AI to help with voice recognition," Gruen said.
AI voice recognition will soon be used to record conversations between patients and their doctor that will automatically transcribe the conversation into the abbreviated language used in physician reports to save large amounts of time and allow physicians to face and discuss care with a patient rather than clicking or filling in EMR boxes. AI also will offer next steps and suggestions based on clinical guidelines, he added.
"So much of what we do as clinicians is still arbitrary, like which antihypertensives or antidepressants we put a patient on. Or for me as a radiologist, what is the right contrast agent for a patient, what is the right dose, what level of radiation do they need, and which protocol should be we use ... and AI is going to be involved in every bit of that," Gruen said.
However, for AI to be helpful rather than hindering physicians and technologists, it needs to be seamlessly integrated into workflows and operating in the background without any additional steps.
"The AI can tell me what to read next, show me what I am missing, tell me which patients can wait, and tell me which patient has an abnormal mammogram right after their exam so they don't have to be recalled. I can spend the time to look at the images now and not have to recall the patient for another appointment, and get them plugged in for their procedure today rather than waiting," Gruen said. "All of this is at out fingertips with AI in the workflow."
Healthcare is moving towards the cloud
Another big trend seen on the floor at HIMSS was the continued movement toward storing data in the cloud, rather than on-site archive storage servers. Vendors said there is no one way to use the cloud. Some hospitals only deploy it for archiving data and still use on-premise servers for current studies and reports. Other health systems are moving completely to the cloud, while some deploy it solely for recovering from ransomware attacks or other disasters.
"I think the movement to the cloud is inevitable. When traveling to hospitals I am asked why do we need to go to the cloud, and why should be go to it today, and we have those answers," Gruen said.
He outlined several reasons to move data to the cloud:
• Large increases in the amount of data being produced in recent years.
• Large increases in the use of AI solutions.
• Cloud helps reduce or eliminate the cost of maintaining on-site servers.
• Cloud can increase cybersecurity.
• Interoperability may be better enabled by using the cloud.
"Having some of you data secured in the cloud is to everybody's advantage," Gruen said. "Whether you are big or small, there is a model for it."
But the biggest benefit of cloud is having immediate access to patient data, regardless of the location of the patient or clinician asking for it. In radiology, there is no reason in this age of technology to still give CDs with imaging studies to patients to bring to their doctor, he said. The cloud enables immediate access to these exams without delay or technical issues that are common with CDs. Pulling prior exams or patient history also can be done rapidly using the cloud.
Interoperability and vendor neutrality is a major trend in health IT
In terms of over-arching health IT trends, Gruen said enabling interoperability and vendor neutrality are major topics. Gone are the days when most PACS and other IT systems were programmed using all proprietary code so it was very difficult to get data from different systems, or data acquired on different vendors' machines, to work where you needed it. Vendors that cannot offer interoperability between the EMR, HIS and other IT systems do not have much of a future in today's healthcare ecosystem.
"People can invest all the AI solutions they want, but we need to be able to present them in a usable way. We know people have ultrasound machines and MRIs from different vendors, and we need to be able to present that data in a usable way. In larger health systems, we also know people may have acquired data in different PACS systems and we need to be able to store them in an archive in a universal way. So we really believe interoperability is good for providers and good for patients," Gruen said.