In pulmonology, AI helps personalize complex procedures

Lung diseases are some of the most common and deadly conditions in the United States and around the globe, affecting hundreds of millions of people worldwide. To reduce the burdens of diseases like asthma, COPD, and lung cancer, pulmonologists are working closely with their clinical colleagues and partners in digital development to make treatments for these conditions more accessible, affordable, and effective. 

And they are increasingly turning to artificial intelligence (AI) to help. AI algorithms are becoming essential for getting more granular insight into the lungs: one of the most complex organs in the body. Without appropriate guidance, largely obtained from imaging studies, sending bronchoscopes or other instruments down the delicate pathways could result in unintentional injury or ineffective treatments. 

“Lungs are incredibly variable,” explained Jean-Paul Charbonnier, CIO at Thirona, which offers AI-driven lung image analysis to providers around the world. “About 70% of people have anatomy that is different than what is currently defined as the standard. That means we need a very personalized approach to identifying those differences and mapping them out so that we can move around inside the lungs safely and treat conditions effectively.” 

“That’s where the synergy between AI and humans comes in. We’re very good at making the qualitative judgements, but AI is much better at the quantitative side of things. The goal is to empower each other so we can do better, do more, and flourish in our work.” 

Leveraging AI for assistance with real-world clinical decisions

Artificial intelligence is maturing incredibly rapidly, bringing new use cases within the reach of front-line providers on a nearly daily basis.   

In pulmonology, one of the main goals is to deploy AI tools that can actively assist with some of the most challenging aspects of the specialty, such as better visualizing the details of the lungs to make more accurate treatment decisions, explains Dirk-Jan Slebos, MD, Head of Pulmonary Medicine and Tuberculosis and a Professor of Pulmonology at University Medical Center Groningen in the Netherlands. 

“AI is being applied successfully in all sorts of scenarios where we need summaries of large data sets, whether that’s hundreds of lines of text in a clinical note or millions of pixels in a radiology image. But where we really want to go is toward the realm of decision support,” he said.  

“In order for AI to provide something truly meaningful to help with treatment decisions, such placing an endobronchial valve, I need more than just the identification that the patient has COPD. I need to understand the exact amount of tissue loss in a certain area of the lung, the exact anatomy of the lung sizes down to a few millimeters, vascular density, and percentages – all of which need to be calculated by AI, because they can’t be seen very reliably by humans.” 

For the past ten years, Thirona has been supporting Pulmonx in building and refining their StratX platform, which enables physicians to upload a CT scan for a patient and quickly receive a report with the clinical data elements they need to assess the patient’s eligibility and prepare for the procedure, in order to accurately and safely perform similar interventions.

“AI allows a much more detailed view into the aspects of the lungs that will help guide my decisions around products and procedures to use for each individual,” said Slebos. “This is the type of approach that will help us make the most out of AI, because it allows us to devote our human effort in the places it’s needed the most.”   

“If we can use AI to learn so much more about our patients during one simple and accessible test, like a CT, we can make it faster and easier for patients to get the care they need while avoiding some of the spending that the US and other places have been struggling with so much.”   

Balancing practicality with openness and innovation during AI development

No one quite knows where AI is going to take the healthcare industry, although most experts agree that it isn’t going to fully replace the role of providers, even in image-heavy specialties like radiology.  Instead, it will continue to augment existing skills and close care gaps that cannot be filled by a dwindling number of clinical professionals. 

However, with the maturity curve and the hype cycle experiencing some misalignment, both providers and developers will need to be careful about offering and adopting algorithms that can truly support better care in the existing clinical environment. 

“There are so many ideas coming from so many different places in academia, in clinical practice, and in the digital health developer community,” observed Charbonnier. “The challenge is that many of these tools are three steps ahead of what the market can actually integrate at the moment.”   

“That’s why we need to stay grounded in clinical practice and work with clinical leaders, like Dr. Slebos, who have that first-hand understanding of what can and needs to be done in the current environment. It’s so important to stay grounded in reality – while maintaining the excitement and possibilities of our long-term visions – so that we can execute correctly on the things that are going to be helpful now while establishing a strong foundation for the future as AI develops.” 

For providers on the front lines of patient care who are uncertain about how AI will affect their day-to-day activities, the advice is simple. “Be open to innovation,” stressed Slebos. “AI can only help us if we’re willing to let it. Let go of some of the old habits and assumptions that we all picked up in medical school or our early days as providers. We’re in a different world now where things have to change, and AI is going to be a major part of that.” 

“It’s going to change the way patients move around the healthcare system and what they expect in terms of personalization and experiences. We have to be open to adapting to those new workflows, relationships, and processes for making decisions that will ultimately support better outcomes for people with lung conditions, because that is the shared goal of us all.” 

Jennifer Bresnick is a journalist and freelance content creator with a decade of experience in the health IT industry.  Her work has focused on leveraging innovative technology tools to create value, improve health equity, and achieve the promises of the learning health system.  She can be reached at jennifer@inklesscreative.com.