From the January/February 2021 issue of HealthCare Business News magazine
Artificial intelligence (AI) has captured the imagination of many people in the five (5) or so years it has been actively promoted. Nearly $1.5 B has been invested in medical imaging AI over the past five or so years, although in fairness, nearly one-third of this went to one company in the cardiovascular marketplace alone. Still, the remaining $1B has its share of companies getting individual and repeat investments of $25M, $50M, $75M or more. Interestingly few, if any, have shown any return on investment (ROI) to date, and based on the slow market growth it will be a while before the venture capitalists start to see a solid ROI as well.
A few months ago, CMS (the Center for Medicare and Medicaid Services) decided to allow payments for AI stroke protocols through a program called NTAP (New Technology Add-On Program). So far, though, only two of the 150-plus vendors in the AI market have been certified to receive this payment. And the rest? They are pretty much on their own to get creative in showing an ROI.
Is AI still of interest? Despite the market shooting itself in the foot early on promoting AI as better than radiologists it’s still a hotbed of interest. Can the cost of AI be justified? Not easily. NTAP works for AI because one of the three criteria for receiving the NTAP payment is “the service or technology must be costly, such that the DRG rate otherwise applicable to discharges involving the medical service or technology is determined to be inadequate”. Using AI isn’t cheap, but it can be very effective.
AI got a lot of interest in the early stages of the COVID-19 pandemic as well with over a dozen vendors developing algorithms for use with chest X-rays (CXR), CT and ultrasound protocols. Sadly, there has been been limited demand for the algorithms, at least in the U.S. Why? The algorithms were never promoted properly and the cost of CT was prohibitive except in the most extreme cases. By the time the AI protocols were recognized by the medical community the development of low-cost rapid (antigen) AI testing had taken away any opportunity that AI had in diagnostic COVID testing. Simply put, imaging had its chances in the early days of COVID when only the PCR (polymerase chain reaction) testing was available. PCR was much more accurate than the RT-PCR testing but could take up to a week to get the results back, whereas a CXR with AI can give them an answer in minutes. This was comparable to the rapid test PCR test, yet rapid tests didn’t gain acceptance until late October/early November 2020. AI simply failed to capitalize on it. It’s still not too late, but with the vaccines finally out now the hope is that the need for testing will slow down and we will finally get a grip on the pandemic.
