Moving closer to commercializing its Watson artificial intelligence technology in imaging applications, IBM announced on Wednesday that it's teaming up with a broad spectrum of partners to help train, test, develop, and deploy Watson in a wide range of organ systems and disease states.
Called the Watson Health Medical Imaging Collaborative, the initiative currently comprises 16 members and includes imaging equipment and software vendors, health systems, academic medical centers, and ambulatory radiology service providers. The collaboration will delve into areas including cancer, diabetes, eye health, brain disease, and heart disease and related conditions, such as stroke, according to the vendor.
"We have done an enormous amount of research in this area, but no one company can do it alone," said imaging industry Anne Le Grand, IBM's new vice president of imaging for Watson Health. "This is [behind] the whole concept of creating a collaborative effort to really make some major breakthroughs in a shorter period of time so that we can more rapidly commercialize relevant solutions to the marketplace."
Anne Le Grand, vice president of imaging for Watson Health.The initial members of the collaborative include Anne Arundel Medical Center in Annapolis, MD; Baptist Health South Florida in Miami; Eastern Virginia Medical School in Norfolk; Radiology Associates of South Florida in Miami; nonprofit health system Sentara Healthcare; teleradiology services provider Sheridan Healthcare; UC San Diego Health; the University of Miami Health System; the University of Vermont Health Network; and teleradiology services provider vRad. Participating vendors include Agfa HealthCare, Hologic, ophthalmology electronic medical record (EMR) firm Ifa Systems, eye-care technology developer Inoveon, ophthalmology equipment firm Topcon, and IBM company Merge Healthcare.
The participants will seek to make use of Watson's ability to gain insights from unstructured imaging data and combine that with information from structured information sources, such as electronic health record (EHR) data and radiology and pathology reports, according to the vendor. The collaboration will take a number of forms. Initially, Watson will be trained in a variety of patient care environments and collaborative members will evaluate potential new offerings, the company said.
Also, participants could help IBM determine how to integrate Watson into existing health IT systems, such as EMR software and PACS. Collaborative members will also work with Watson Health to train Watson for various health conditions using their own deidentified data or data from population-based disease registries. Meanwhile, vendor members could also integrate Watson technology into their workflow or image management software, according to the firm.
Focused collaborations
Collaborations will revolve around particular areas, such as forms of cancer, brain disease, or stroke. For example, an academic partner or partners that may have performed a significant amount of breast cancer research would work with a commercial partner or partners with a similar focus in developing products in that area, Le Grand said.
"So you might have the academic partner provide the unstructured/structured data and original, deidentified patient record information, and commercial partners will create the solutions," she said. "The component that Watson Health Imaging adds is the cognitive capability to take the information, utilize it, and give meaningful data back so that the solution is more relevant. And because it's cognitive it will continue to learn."
Community hospitals will also be involved in the process, said Steve Tolle, IBM Watson Health Imaging's chief strategy officer.
"We have to make sure that all of the buyers in the market can see themselves in this work," Tolle said. "It depends on the organ system or the disease that they're going after, but [the collaborations] will typically be a mix of an academic and community hospital and potentially an IT partner as well."
Not every partner will work on every disease or target area, he noted.
"We will be dividing them up across the entire body," Tolle said.
However, "there will also be some very core technologies and information that is shared across" the collaborative, according to Le Grand.
In cancer, the collaboration will initially focus on the breast. For example, physicians could make use of Watson predictions when diagnosing their patients, she said. In addition, Watson could be used to review images retrospectively as a way of improving accuracy for a particular clinician, pointing out certain areas of interest.
Some of the collaborative participants would also like to use Watson to help develop new breast screening protocols based on factors such as patient age or breast tissue density, Tolle said.
The lung will be a focus area in cancer in the future, and the prostate is also being considered as well, Le Grand said.
In addition to cancers, the collaborative will focus on applying Watson in areas such as diabetes, eye health, brain disease, and heart disease and related conditions, such as stroke.
Commercialization plans
Capping off more than 10 years of research, IBM expects to bring its Watson product -- quality improvement software for identifying patients who are candidates for care management intervention -- to market by the end of this year or early next year.
"We're starting in the clinical area of aortic stenosis, which happens to be where IBM has done a lot of research around valve disease and the heart," Tolle said. "Then we will add other clinical indications to that platform over time; breast imaging will be next, and then you'll see us do some work in the lungs."
In early 2017, IBM also hopes to roll out an EMR summary service, Tolle said.
"Evidence that has been published in the scientific journals shows that when a radiologist has access to a patient's medical record, they are likely to change their decision or recommendation about 18% of the time," Tolle said. "We will be introducing a service that will read a medical record as well as progress notes, structure it, and provide a relevant summary to the diagnosing physician, which will be a radiologist or cardiologist."
The service could also be used to provide access to these summaries for physicians at the bedside, Le Grand said.
In the future, IBM expects to add more members to the collaborative, Tolle said.
"We're starting with the human body and organ system and in time we'll then go to kids," he said. "At some point we may get into fetal medicine, so [there's] lots of opportunity to collaborate with many different players at different levels of experience."
In terms of a regulatory pathway for these technologies, IBM will work closely with the U.S. Food and Drug Administration to determine the appropriate course of action based on claims that will be made for these products, Le Grand said.
