VIDEO: Development of AI app stores to enable easier access

Dreyer also holds the positions of vice chairman of radiology at Massachusetts General Hospital, chief data science and information officer for the departments of radiology for both Massachusetts General Hospital and Brigham and Women's Hospital, and associate professor of radiology at the Harvard Medical School. 

The development of medical imaging AI algorithms has been very piecemeal with apps developed by numerous, separate companies. This requires hospitals or health systems to contract with several small companies to access these applications, which can be a nightmare for hospital informatics departments. This led several larger radiology software companies to adopt a type of app-store approach, offering several AI products from numerous vendors in one location that meet some basic integration criteria.

That was the trend in how to access AI as of 2019, but in the two years since COVID hit, that trend appeared to be altered at the HIMSS and RSNA 2021 meetings. The trend there appeared to be transition by several radiology vendors from the app store concept to full partnerships with key AI developers to integrate their algorithms directly into their PACS or enterprise imaging systems. This allows for a seamless interface and the AI piece then becomes just another tool in the options submenus without requiring another login or separate report or storage of the AI assessed images. 

Dreyer said it makes sense for PACS providers to integrate AI tool directly into their specific systems, but for health systems that do not use these vendor's systems, there is still a need for an "app store" like market. He believes that model will be the way of the future, as long as the AI can integrate easily into existing PACS with industry standards for health informatics.

"There are more than 200 AI algorithms and it seems to be following Moore's Law,": Dreyer said."But it is going to be unmanageable to deploy 20 or 30 algorithms from 20 more 30 different vendors."

He said this is a similar situation that existed with MP3 players and music download services in the early 2000s, and it was unmanageable to download all these pieces onto a hard drive. But, the industry rapidly changed and adopted the technology when it was standardized by Apple and its iPod and Apple Store concept. He said something similar will be needed in radiology AI.

"I really feel that is the only way that a large part of AI is going to work. The other part of that is embedded AI. I do thing that PACS systems will have embedded AI, EHR systems will have embedded AI, imaging modalities will have their own AI. But if you want independent developers of AI to solve problems in the diagnostic cockpit of radiology, it at some point has to go through one or two app stores." 

He said that is a beauty of the Apple Store concept, is that independent software developers can create their own apps to industry standards and then easily and make them available through a very public outlet. That is what he said is needed for the future of AI algorithms for healthcare.

This is part of a 4 video interview series with Dreyer on various aspects of radiology AI. Watch the others:

VIDEO: Where will radiology AI be in 5 years?

VIDEO: Overview of radiology AI by Keith Dreyer

VIDEO: Segmenting the Radiology Artificial Intelligence Market by Function