General Electric’s healthcare division will partner with the corporate parent of two of Harvard University’s teaching hospitals to develop artificial intelligence products for medicine. The goal: to leverage the company’s dominant position in medical imaging into a new ownership of medical AI.
“We make machines but that’s not really the business,” says John Flannery, the chief executive of GE Healthcare, says he tells his team. “The business is what kind of solution can we put together that gets a better clinical and economic outcome?”
GE Healthcare is one of the main manufacturers of imaging devices used in radiology–things like PET and MRI scanners. The company also sells the computer software that doctors use to manage the images. This area, long thought to be low-hanging fruit for artificial intelligence applications, could form the basis of the first products to emerge from the collaboration. For instance, software might be able to identify which scans are normal, freeing human radiologists to focus on cases where there is an abnormality. (In 2015, IBM bought a company focused on reading radiology images with similar aims.)
The deal is between Partners Healthcare, the corporate parent of the Massachusetts General Hospital and the Brigham & Women’s Hospital, two Harvard-associated teaching hospitals.
“We’ve committed within our organizations to have a center that’s going to endure and the relationship with GE is intentionally 10 years,” Dr. Keith Dreyer, Chief Data Science Officer at MGH and BWH. “This is going to take a while to do. People ask me how long this is going to take to happen. And I say how long is it going to take for the Internet to happen?”
Dreyer says that the field of using AI in medicine has gone from hundreds of scientific publications to thousands over the past three years. He contrasts the effort to create an AI that would serve, essentially, as a virtual physician to a more realistic approach that is piecemeal, dealing with one opportunity at a time.
Flannery, the GE Healthcare CEO, is eager to move fast. He says commercialization of cloud-based radiology applications could occur in one to three years. “There will be an iTunes-like app store of programs offered to the physicians,” Flannery says. Other companies would then use the infrastructure and data created by GE and its partners to create their own programs, all of which will run on a single application. “If there were 20 technologies with these solutions, I can’t buy 20 solutions. I need a single platform,” Dreyer says. This remains an interesting space to watch.
Full article: https://www.forbes.com/sites/matthewherper/2017/05/17/ge-partners-with-harvard-hospitals-to-put-artificial-intelligence-in-medicine/#30e8838b7385