Artificial intelligence-powered medical imaging analysis startup Aidoc Medical Ltd. believes it can play a key role in helping hospitals implement their AI strategies with today’s launch of what it says is the first AI Operating System for the healthcare industry.
The new platform is designed to enable the clinical use of AI applications at scale using a unified system with better orchestration capabilities.
Aidoc is applying advanced AI to the area of radiology to help clinicians make faster and more accurate diagnoses of a range of medical issues. The company has built a range of AI models that help radiologists to analyze computerized tomography or CT scans in real-time. Its “always-on” technology runs behind the scenes to identify urgent cases where immediate medical intervention can possibly save someone’s life.
Aidoc’s AI models, which are pre-trained on more than a million CT scan images, work by trying to mimic the way a human radiologist would analyze them. The models learn from their past experience, which enables them to improve over time and become more accurate.
The promise of Aidoc’s AI is not lost on hospitals. A survey earlier this year by Sage Growth Partners revealed that 90% of U.S. hospitals have an AI strategy in place, up from just 53% in 2019. The problem lies in implementing those strategies, with just 34% of hospitals having actually deployed an AI system.
Aidoc reckons this low rate of deployment is the result of the lack of a “unifying quality” necessary to integrate and orchestrate multiple AI solutions at large scale, due in part to limitations caused by vendor incompatibilities.
It’s this challenge Aidoc is looking to solve with its new AI Operating System, which integrates not only the startup’s own AI tools but also those of many third-party providers. The idea is it can be used to orchestrate a diverse set of AI models under a single, unified operating system, using AI-powered image analysis to determine the most relevant algorithm for each CT scan.
Dr. Paul Chang, vice-chair of Radiology Informatics at the University of Chicago, said it’s not enough simply to provide hospitals with a selection of plug-and-play AI solutions.
“A true platform must be based on a unified architecture that enables effective use of various AI products in the challenging real-life IT settings of health systems,” he said. “That is exactly what Aidoc’s AI OS solves.”
Aidoc offers seven Federal Drug Administration-cleared AI models that can be used to analyze CT scans for intracranial hemorrhage, acute C-spine fractures, intra-abdominal free gas, rib fractures, large vessel occlusions and pulmonary embolism. In addition to those, Aiddoc’s AI Operating System integrates with AI models from companies such as Imbio LLC, Riverain Technologies Inc., Icometrix NV, Subtle Medical Inc. and ScreenPoint Medical BV to cover a wide range of radiology subspecialties and imaging enhancements.
Aidoc Chief Technology Officer Michael Braginsky said his company came to realize the challenges of implementing AI and the need for a unified system when trying to deploy its own AI modules for its customers. He said the company was forced to create an AI operating system that can run in any clinical setting, and that it made sense for it to play nicely with other services.
“We did so with one question in mind, ‘How do we bring value to the providers?’,” Braginsky said. “More and more companies who’ve developed amazing solutions are operating on our OS, and we are proud to enable them to greatly impact patient care.”
Analyst Holger Mueller of Constellation Research Inc. said AI is expanding across more and more lines of business and that it has a lot of promise in the healthcare industry, especially in the image processing space. That’s because computers can process images thousands of times faster than humans will ever be to do, he said.
“There is a small infrastructure problem that needs to be overcome, though, so it’s good to see an effort like this from Aidoc with its new operating system helping to address orchestration,” Mueller continued. “It’ll be interesting to see how successful the platform is, as the healthcare industry’s adoption cycles are often painstakingly slow.”