It's clear that AI will have an impact on the industry, but how should you be investing now?
If you think back over the hot topics in radiology over the past few years, one subject frequently rises to the top of the list-artificial intelligence (AI). Much publication space and conference time have been devoted to discussing how the advent of this technology will impact-or is already impacting-daily practice and patient care.
But, one significant question remains-how will these new tools affect the way you make purchasing plans and decisions?
Related article: Artificial Intelligence: Own it Now
According to industry experts, it’s too early to know how AI will impact these financial choices in a concrete, rubber-meets-road way. However, having a better understanding of how to recognize the most promising types of tools will help you make informed technology selections in the future.
“We’re so early in the process with artificial intelligence, that it’s still very much a black box,” says Lauren Golding, MD, a radiologist with Triad Radiology Associates in North Carolina. “No one knows what the artificial intelligence tools are going to look like, so it’s hard to know how to value them and how to think about them within a reimbursement structure.”
Until the industry has a clearer picture of the AI tools that will be available, such as software that could be used to make identifications on every study or back-end processing, she says, actionable guidance about how to create purchasing plans or to make individual buying decisions will be elusive.
The AI frenzy
It’s clear that that the healthcare industry, including radiology, is set to embrace AI technology as it advances. Based on a recent report from Signify Research, an independent global healthcare market intelligence and consultancy firm, medical imaging tools that facilitate automated detection, quantification, decision support, and diagnosis are set to become a $2-billion industry by 2023.
Despite the buzz and high expectations, though, the industry has a long way to go for AI to become as effective and influential as many have been led to believe it already is. The intense excitement and hand-wringing around AI has created what Hugh Harvey, MD, radiologist and clinical director at Kheiron Medical in London, dubs the hype curve.
“For all the dreams and start-up bluster, there’s a mountain of hard work, regulatory bureaucracy, scientific validation, and institutional bias to overcome,” he says. “None of this is insurmountable, but to stand the best chance of succeeding, we need to take a step back and survey our surroundings before deciding on the best investment strategy.”
What you need to know
If you’re looking to purchase any AI tools now-such as software that can triage head CT scans that include head bleeds-you should consider your reimbursement limitations, says Golding, faculty for the 2018 American College of Radiology Annual Conference on Quality and Safety.
In a fee-for-service environment, she said, it isn’t clear how you’ll be reimbursed for using the tools. Radiologist-owned practices that own their own imaging centers could only be reimbursed through the physician fee schedule if a CPT code were created for a specific AI tool and it was valued through the RVU system.
“It would have to be classified as physician work, and there’s not a lot of physician work with AI tools,” she says, noting purchasing such tools under this system could weaken your bottom line. “So, it’s hard to quantify in this paradigm. It’s just unlikely in fee-for-service that we’re going to see meaningful reimbursement for artificial intelligence tools.”
Related article: Artificial Intelligence Makes Gains in Radiology
This will change as the industry continues to move more toward value-based reimbursement, though, she says. The other option pushing for greater AI tool adoption, she says, could be a government mandate that practices increase implementation in order to comply with Meaningful Use.
As either of these scenarios approaches fruition, Harvey says, there will be several things for you to consider. Here are six tips he has for deciding which AI tools or companies are worth integrating into your practice:
1. Focus on specifics: Choose tools that focus on specific medical problems, avoiding ones that claim to offer automated solutions for every diagnostic or workflow activity for a given modality. Opt for tools backed up by proven clinical results.
2. Play by the rules: Purchase AI tools from companies that are familiar with all the burgeoning regulations that will govern how you use the technology, he says. If not, you could find yourself with software that isn’t certified and approved for clinical use, undermining your expenditure.
3. Don’t try to remove people: Do not purchase an AI tool that markets itself as a radiologist-free device or system. The FDA will only approve automated diagnostic and detection tools that operate under human observation. Instead, select tools designed to augment your radiological contributions, such as triage systems, quantitative analysis tools, and registration or segmentation systems.
Related article: Artificial intelligence in radiology: Friend or foe?
4. Choose tools that meet a real need: Avoid spending money on AI tools that do things you really don’t need. Ask yourself and your colleagues if there’s a cheaper or easier way to complete a task, and, if so, forego the purchase, Harvey says. Identify your clinical problems and needs before you search for automated assistance. That way, you’re confident you’re spending capital on a device, software, or system that will help you add value.
5. Ask about bias: AI tools can’t be trained to catch every finding, and they’re only as good as the clinical data on which they are taught, so they’re inherently vulnerable to bias. However, investigate which companies and tools have taken steps to counteract any algorithm biases. Selecting a tool that is precise will augment your level of patient care, enhancing the value of your purchase.
6. Think big and bigger: At the grassroots patient care level, selecting AI tools proficient in image interpretation will help speed up your diagnostics and, potentially, eliminate the need for unnecessary follow-up exams. But, Harvey says, look to embrace the full measure of AI’s capabilities. When the time comes, consider sinking capital into solutions that anonymize medical data, natural language processing for structuring clinical free text, and application programming indices that support health information exchange interoperability and access to electronic health record data. Purchasing these tools when-and if-they become available could put you at the forefront of leveraging AI to maximize your value within your health system.
Even with these guidelines, Golding cautions that many AI tools available on the market today are not sophisticated enough to meet your clinical needs. In fact, the vast majority are still being vetted in academic environments, but the developments are worth watching closely to ensure you make informed future purchasing decisions.
Related article: AI and the Future of Radiology
“We always need to be aware of what’s going on. It’s important to know all the policies and regulations around new things, and AI is no different even though it’s not particularly developed right now,” she says. “I think it can safely be watched at a distance, and we can see what unfolds over the next year or two when there might actually be tools out there that have specific uses. You need to think about how you can improve your practice, efficiency, and quality of care so when a tool comes along that fits in that area, you can latch on to it and go with it.”
Mammography News: FDA Grants Expanded 510(k) Clearance for AI-Powered SmartMammo Dx DBT Software
November 29th 2024Originally cleared by the FDA in 2021, the SmartMammo Dx software for digital breast tomosynthesis (DBT) can now be utilized with the Senographe Pristina mammography systems from GE HealthCare.
The Reading Room Podcast: Emerging Trends in the Radiology Workforce
February 11th 2022Richard Duszak, MD, and Mina Makary, MD, discuss a number of issues, ranging from demographic trends and NPRPs to physician burnout and medical student recruitment, that figure to impact the radiology workforce now and in the near future.
FDA Clears AI-Powered Ultrasound Software for Cardiac Amyloidosis Detection
November 20th 2024The AI-enabled EchoGo® Amyloidosis software for echocardiography has reportedly demonstrated an 84.5 percent sensitivity rate for diagnosing cardiac amyloidosis in heart failure patients 65 years of age and older.