THANK YOU FOR SUBSCRIBING
Over the past 20 years, technology has helped radiologists significantly improve efficiency which, in turn, has also improved care delivery. Picture archiving and communication systems (PACS) have enabled us to look at imaging studies digitally on a computer instead of having to read off films, boards and scopes. Instead of spending hours per day hanging films, we can not only look at patient studies but also see any prior studies of those patients much quicker. Where we used to use transcriptionists to transcribe our reports, which would be signed and sent out 24 hours later, today we use voice recognition technology that lets referring physicians get the reports significantly faster– sometimes within just minutes–and take action much sooner on patients who need it.
The future promises to be far more transformative, as we explore how to use technology to make us better physicians.
"Reducing U.S. radiologist errors by just 1 percent could improve or save the lives of hundreds of thousands of patients per year"
Current Challenges and Limitations
Probably the biggest challenge radiologists face today is that we don’t have all the patient information we need to make our readings more accurate. Referring physicians rarely provide all the relevant patient information; the most common “reason for exam” on the order sheet is “pain.” A lack of system interoperability often prevents us from accessing the information via the patient’s EMR (electronic medical record). For example, a radiologist in Florida who is reviewing the study of a patient who was seen in California may not have access to that patient’s medical information and history because the two EMR systems can’t talk to each other. Similarly, if radiologists read on different PACS systems, lack of interoperability can limit a physician’s ability to tap into a particular radiologist’s expertise. At Sheridan Healthcare we have a group of 350 radiologists who specialize in different areas and subspecialties, but the fact that they don’t all read on the same PACS limits our access to them. We’re now in the process of linking them all together, which will give us access to all 350 radiologists and allow us to choose the right radiologist to read a particular study. For example, we’ll be able to have a board-certified neuroradiologist, rather than a general radiologist or an abdominal radiologist, read a complex MRI of a patient’s brain.
Using the Power of IBM Watson to Transform Care
IBM’s recently formed Watson Health medical imaging collaborative comprises 16 foundational members, including Sheridan Healthcare. The partners– leading health systems, academic medical centers, ambulatory radiology providers and imaging technology companies–will be leveraging Watson’s capability to “read” and understand previously “invisible” unstructured imaging data, teaching it to extract insights and combine them with a broad variety of data from other sources, such as data from electronic health records, radiology and pathology reports, lab results, doctors’ progress notes, medical journals, clinical care guidelines and published outcomes studies. Our ultimate goal is to take advantage of Watson’s unparalleled cognitive computing capabilities to help doctors make personalized care decisions about individual patients, simultaneously building a body of knowledge that will benefit broader patient populations.
Using IBM’s IASO and Avicenna software, which can read, integrate, understand information from text, structured and unstructured data, and medical imaging, we will be teaching Watson how to review a patient’s medical record and present the relevant information for a particular exam, helping to ensure that radiologists will have all the relevant information they need to perform more accurate readings. Avicenna’s “reasoning” system is even being taught to suggest possible diagnoses.
I think the most exciting aspect of this collaborative effort is the potential for Watson to help reduce human errors. Any radiologist, no matter how experienced, knowledgeable and detail-oriented, is likely to miss something at some point. We will be working to teach Watson to bring things radiologists might otherwise have missed to their attention and, in the process, help them learn to look at studies differently and read them more accurately.
A big part of Sheridan’s role in the collaborative will be helping IBM determine the optimal use cases to focus on. We are basing those use cases on a combination of practical considerations– fairly clear-cut things that a computer can be taught to identify reliably and what we think radiologists are likely to use on a day-to-day basis–and what has the potential to make the biggest impact on patient care. One of the use cases we’ve identified for reducing human errors is fractures, which can be very subtle and therefore frequently go undiagnosed. Those missed diagnoses can have a major negative impact on the patient. For example, if a hip fracture in an elderly woman goes undiagnosed and she slips and falls the next day, the fracture might cause blood clots, pulmonary emboli or other major ill effects could cause her to be extremely sick and disabled for many years, or even kill her. If we can be more accurate in detecting those fractures upstream, we can prevent a lot of morbidity and mortality downstream. There are other subtle findings that might be missed by even the best radiologist. If Watson can help detect these subtle abnormalities, that finding could prevent the patient from being sent home and, potentially, having a catastrophic event there without access to emergent care.
Prioritizing cases that require emergent care is another challenge. We’re hoping to teach Watson to sift through a stack of films, reliably identify potentially critical or life-threatening problems and prioritize those studies at the top of a radiologist’s work list. If Watson could identify a critical finding and prioritize it appropriately, it might cut the patient’s time to emergent care delivery by 30-60 minutes–potentially the difference between life and death.
Reducing U.S. radiologist errors by just 1 percent could improve or save the lives of hundreds of thousands of patients per year. The collective technical and medical expertise of the Watson Health medical imaging collaborative, together with IBM’s annual research and development budget of more than $6 billion, may allow us to turn this vision into reality very soon. I’m tremendously excited about Sheridan Radiology being a part of this effort, and even more excited about the impact of this collaboration between humans and technology to improve the lives of the patients we care for every day.