As every corner of the healthcare industry prepares to leverage the power of AI for clinical and operational improvements, there is an obvious role for the growing ecosystem of AI developers. By some estimates, there are more than 200 AI-based medical imaging startups.

To build healthcare AI tools at speed and scale – and so streamline the commercialization of these promising, assistive tools – partners need a strong, experienced vendor who can support their needs and go-to-market goals.

“We need to partner with innovators to complement our own offerings,” said Abby Pandya, Director of Product Management for AI & Analytics Ecosystem at GE Healthcare. Partners are needed, she said, because “we can’t create all the solutions our customer will want and need.”

The GE Healthcare Edison Developer Program, announced at the Radiological Society of North America 105th Scientific Assembly in December 2019, helps market-ready independent software vendors and builders deploy and deliver advanced healthcare applications, integrate products and drive commercial growth worldwide. Paired with GE’s own apps marketplace (apps.gehealthcare.com), the developer program benefits healthcare organizations and doctors, too. It simplifies the discovery, evaluation, purchase and deployment of many AI algorithms and solutions from many developers.

From a technology perspective, GE’s approach is very intentional. “A key principle is that these solutions are not one-off tools but are integrated into the customer’s existing workflow,” Pandya said. Specifically, GE Healthcare provides configuration set-up guidelines and APIs to ensure a developer’s product is compatible with the target integration product. This applies to solutions on the device, in the PACS or in the cloud.

From a business perspective, GE equips its Developer Partner Program participants with various support services, including marketing and sales. This is intended to help an innovative developer get over barriers to adoption. Developer participants gain access to GE Healthcare’s install base of 4 million imaging, mobile diagnostic and monitoring units in over 160 countries.

Developers in the GE Healthcare Edison Developer Program leverage three “value themes”:

• Interface. “Incorporate your application with the standards-based interfaces on the Edison platform to streamline the commercialization of your technology for faster adoption – whether you’re on device, on premise or on cloud.”
• Grow. “Accessing GE Healthcare’s global channel allows you to scale clinical and commercial activation, thanks to the coverage enjoyed across our expansive install base.”
• Trust. “Partnering with GE Healthcare ensures you’re joining a global community you can trust, guided by AI principles and a century-long track record of delivering healthcare innovation.”^[1]

Example one: Koios Medical

For example, Koios Medical is the only FDA-cleared AI software for the analysis and classification of breast lesions in ultrasound exams. Koios partnered with GE Healthcare to incorporate their solution with GE Healthcare’s LOGIQ E10 mobile scanner.

“Koios DS is not a detection algorithm but rather is decision support,” said Koios President and CEO Chad McClennan. “There’s literally a Koios button on the LOGIQE10 console that activates analysis on an image in two seconds or less and presents a recommendation that can then be exported to the clinical record.” The pixel-level analysis is based on a database that exceeds 500,000 ultrasound images. Breast Assistant, powered by Koios DS^TM, which is available as an upgrade, is compatible with LOGIQ E10 and GE Healthcare’s Centricity Open PACS^[2] and GE Healthcare’s Invenia ABUS (automated breast ultrasound).

Providing this sort of automated “second opinion” decision support for suspect images, in real time, cuts down on avoidable biopsies, said McClennan. He added that without a second opinion, on average, up to four of every five – or 80% – of biopsies come back as benign.

“Our software reduces clinical variability and clusters performance of all users at high levels of clinical accuracy,” he said. “We catch now up to six more cancers per hundred presented, while reducing the number of benign biopsies, on average, by 30%. This is a new paradigm to improve both simultaneously, delivering on the promise of AI in elevating patient care.”

As a developer, Koios worked closely with GE Healthcare engineers on enhancing the user interface (UI) and integrating the solution into the clinical workflow. Specifically, Koios DS expresses its risk assessment findings in terms of the appropriate American College of Radiology’s BI-RADS© category, enabling proper patient management.

Example two: MaxQ AI

Another developer in the program is MaxQ AI, which uses deep learning and machine vision for clinical diagnostic intelligent software solutions that enable timely and more accurate diagnosis of head trauma and stroke.

“The biggest benefit for developers is that collaborative relationship with the GE channel, working with GE Healthcare’s clinical and marketing teams to ensure our solution will resonate in the marketplace,” said Randy Rohmer, MaxQ AI’s Director of Commercial Operations. Rohmer said this relationship also ensures tight integration of the solution into the PACS, a key feature. “If you’re not seamlessly integrated with the PACS workflow, [clinicians] are not going to use the solution,” he said.

In late 2018, MaxQ AI and GE Healthcare announced that MaxQ AI’s ACCIPIO AI platform would be a part of GE Healthcare’s Computed Tomography (CT) Smart Subscription offering. GE Healthcare’s Smart Subscription provides customers with continuous access to the latest imaging software updates through an available automatic download to their CT scanners from the cloud.

In addition to MaxQ AI’s ACCIPIO Ix, a triage tool for intracranial hemorrhage (ICH) detection that is included in each GE Smart Subscription, customers will soon be able to upgrade functionality with ACCIPIO Ax, which annotates images at the device level. (ACCIPIO Ax is expected to gain FDA approval in October.)

Most of the developers in GE’s curated ecosystem today are like MaxQ, with demonstrated regional, regulatory and commercial traction. From the lens of a hospital technology buyer, this robust, pre-vetted ecosystem offers great benefits.

Pandya said there are plans to broaden the ecosystem by further expanding the number of developers. “We’re talking to a lot of these startups, to see who is on that leading edge,” Pandya said. These conversations take place in the context of GE Healthcare’s research into customer needs, technology trends and market reports. The goal is to find developer partners who align with these learnings.

“You’ve got extraordinary access to customers and prospects,” said MaxQ AI’s Rohmer, adding this is invaluable for a company like his, which needs to understand customer priorities for improving efficiency or staff and patient experience.

GE Healthcare’s Edison Developer Program is a year old. In addition to the developer program website at www.gehealthcare.com/edisondeveloper, GE Healthcare has an apps marketplace at apps.gehealthcare.com. The marketplace features both internally developed and third-party applications. Separate from both the developer program and the apps marketplace, GE Healthcare is working with a number of clinical partners, such as academic medical centers, to develop new care-area solutions, which should be market-ready in two or three years.