Edwin Warfield, CEO of citybiz.co, interviews Jan De Backer, founder and CEO of FLUIDDA, a company using artificial intelligence (AI) to help physicians identify and diagnose lung diseases. De Backer discusses threats to lung health, including pollution and long COVID and populations with a higher risk of lung diseases including fire fighters and members of the military. He explains why identifying lung disease is a unique challenge and the limitations of current diagnosis technology. With FLUIDDA, De Backer used what he learned as an aerospace engineer and applied it to the lungs to simulate airflows and detect abnormalities. He explains how leveraging AI enhances physicians’ ability to detect lung disease earlier, treat it more effectively and develop better drugs and bring them to market more efficiently.

Explain the complexity of lung diseases and the current state of pulmonary diagnostics?

We’re just coming out of (or likely still in) a respiratory pandemic, so millions of people have recently experienced the complexities of lung diseases firsthand.

The problem with lung disease, it’s hard to diagnose and often only diagnosed in the very late stages because lungs have a remarkable capability to compensate for diseases. Often, certain parts of the lungs are already affected by diseases (like asthma or COPD caused by smoking), but the way we measure lung function today, does not detect an abnormality.

Advances in imaging are opening a black box and revealing the regional parts of the lung with quantification methods to help medical professionals better understand the true extent of the disease. That’s something that became apparent at the beginning of COVID. We didn’t know what the disease was doing. Why were these people on ventilators dying at such a high rate? We have developed more insights into this over the last couple years.

What are the limitations of conventional CT scans?

The conventional way of assessing a CT scan is having a radiologist visually scroll through the two-dimensional images looking for abnormal patterns. They often only spend three to five minutes looking at a CT before providing the report. It’s usually a subjective measure so five radiologists looking at the same image will probably have slightly different interpretations. It’s also not very scalable and there are certain features that you also cannot see with the naked eye.

As an example, it was obvious in COVID patients that the blood volume is distributed over the different generations of blood vessels. So, blood vessels in the lung operate as a bifurcating system. If you’re healthy it follows a more normal exponential curve. If you have a disease, there are shifts away from that curve. That’s something you can’t see with the with the naked eye because it’s really a 3-D phenomenon. That’s where advanced quantitative methods start playing an increasingly important role, providing information that’s hidden in plain sight. This is not possible to detect just by looking at CT scan.

Can you explain functional respiratory imaging (FRI)?

My background is in aerospace engineering. In that world we use computers to simulate air flows around wings and engines. My father is a respiratory physician and asked if we could use a similar approach to simulate the air flows in the lungs.

One of his jobs was to assess how well lung disease drugs are working, but conventional lung function tests made it difficult to understand what was really going on. So we combined methods from aerospace engineering with the typical CT scans, to develop this FRI technology which does two things: one, provide better insights into lung diseases, so we have a better idea of what part of the lungs are affected; and two, match the right patients with the right treatments.

We now call this technology functional respiratory imaging (FRI). It’s functional because it uses these methods from aerospace engineering looking at flow behavior. It’s respiratory because we focus on the respiratory system and it’s imaging because it uses CT scans, the typical medical imaging modalities.

What is the potential impact of FRI on asthma and other respiratory diseases and what is its impact on drug development?

Whether it’s the effects of smoking, asthma, COVID or exposure to toxins in the air like we’ve seen with servicemen in burn pit environments and firemen, the problem is that it detected too late using conventional CT scans and physician evaluation. FRI allows you to detect abnormalities at an earlier stage.

Lung diseases are generally irreversible. Once you lose lung function, you hardly ever get it back. If you detect it earlier, you can make sure the patient maintains a higher level of lung function for a longer period, which is associated with better quality of life.

The same thing is true for developing drugs. If you can only detect decline in the later stages, you’ll only develop drugs that are targeted towards the late stage of the disease. More sensitive technology enables earlier intervention. Drugs and drug effects can be evaluated in earlier stages of the disease.

This technology can also substantially reduce the sample size and duration of clinical trials. You can study a drug with fewer patients over a shorter period, develop better drugs and bring them to market faster.

What role is machine learning playing in your technology?

We’re using artificial intelligence (AI) and machine learning in several ways. One is to improve the extraction of information from CT scans. A CT scan generally has color labeling to identify airways, blood vessels, fibrosis etc. That’s input for machine and deep learning, artificial intelligence approaches where the machine learns how to do it autonomously.

Instead of having individuals perform manual processing, the machine can progressively take over more duties and perform an accurate job automatically so it’s almost an instantaneous processing. When we started 15 years ago it took about a week to process a patient. Now it’s a matter of minutes.

Machines are also very good at bringing a lot of different types of information together and making assessment. For example, a machine can look at the vasculature, the airways, the long volume of fibrosis and use all that information and make an interpretation assessment.

Humans traditionally look at one element at a time and often lack a more comprehensive picture. Radiologists scroll through the image looking for abnormal patterns and make a diagnosis in three to five minutes without additional data like blood vessel characteristics, characteristics of the airways, etc. That’s where AI and machine learning can enhance the performance of the physician.

To be clear, AI will not replace physicians. But physicians who use AI will probably replace physicians who don’t.

Roughly 300 clinical centers have been trained worldwide, mainly participate in clinical trials. We project that number to go up to about 500 in the next 12 to 18 months.

Please provide some background about your company.

FLUIDDA was started about 15 years ago in Belgium to develop technology that would help improve respiratory treatments and diagnoses. Today, we have offices across the world, including Los Angeles, New York, Atlanta, Lisbon, Portugal, and Singapore. We’re still small and a bit spread out because we’re doing a lot of clinical trials that require us to be close to the hospitals participating in those trials.

What are the growth plans for the company?

There’s tremendous opportunity. We receive a variety of requests from pharmacy and medical device companies to participate in their clinical trials by designing or optimizing their drug evaluations. One big opportunity is making sure that technology is being used efficiently in hospitals, so we target our services to a broad range of hospitals. Academic research centers also see significant value in the technology.

In the cardiopulmonary rehabilitation environments, we’re seeing many people suffering from the lingering consequences of COVID. We see even relatively young people in their 30s that have extreme fatigue. It’s difficult to discern exactly what’s going on because a CT scan provides a radiology report saying everything looks normal. But if we measure these scans with technology, we see that it’s not normal. There are abnormalities that are not apparent to the human eye, but they’re still there.

That’s where the right technology can be useful. It enables physicians to guide the patient to a more optimal cardiopulmonary rehabilitation program. We’ve seen improvements in virtually all patients.

It can also give patients hope. There is a pivotal point where the medical system can provide hope. If healthcare doesn’t provide hope, it often sparks a negative spiral. Patients and their families get depressed and the thought of going to the doctor or even trying to get an appointment can deepen the depression.

In a technology driven healthcare environment, we can provide care and improve clinical outcomes very efficiently. We must turn the tables a little to ensure that we build a system from the bottom up but driven by technology and data. Healthcare professionals working in this environment can leverage a host of tools to improve patient outcomes and increase efficiency.

Future growth and innovation in healthcare will be built from the bottom up in a technology and data driven environment. That’s the future of healthcare and we’re excited to contribute to that future.

You’ve started a new blog that brings surfing and chess. Tell us about the origins of that and tell us about some of your topics that you’ve written about?

A year ago, I was thinking about the last 15 years in business and the characteristics needed as an entrepreneur to succeed in healthcare. I saw that you simultaneously need the perspectives of a surfer and a chess player.

The surfer’s perspective is attuned to the macro environment. Everything in healthcare moves relatively slowly, but if a wave is coming you need to anticipate and feel it to catch it. If you start paddling too soon or too late, you miss the window to catch the wave and must wait for the next one. The wave is not under your control so, as an entrepreneur, you need to feel where the environment is heading in order to be in the right place at the right time.

Once you catch in the wave and gain momentum you begin to have more control of where you go. You must be aware of other surfers and their movements and that’s when the chess player’s perspective is useful, plotting the next steps and strategies that will give you the best chance to advance and ultimately win the game.

That’s the setup for the blog. The goal is to explore and promote innovation in health care that connects the silos and investigate where technology fits into the overall healthcare environment. We also feature interviews with the patients and professionals that benefit from this technology. For instance, last week we interviewed a general from the national guard who developed lung diseases while serving in in Afghanistan. She has been frustrated with the conventional healthcare system because they can’t fully diagnose or treat the issue. She was categorized as having asthma, which is probably partially true, but likely not the whole story.

It’s these stories, especially from those who put their lives and health on the line to serve their country, that show where technology can help and make sure that they either stay stable or hopefully improve. So, one goal of the blog will be to bring those stories to light and put a spotlight on these issues to demonstrate how technology and innovation can improve patient outcomes and also improve the system for healthcare professionals.