Schedule a Meeting

This is Conversations with Digital Diagnostics, a series where we delve into insightful discussions with company thought leaders.

Prior to joining Digital Diagnostics, John held a variety of executive roles at Epic Systems. During his 12-year career with Epic, he led cross-functional teams in a variety of product, customer success, and business development roles with a focus on growth. He continues to be passionate about the application of artificial intelligence and computer vision in healthcare. 

Digital Diagnostics: As an AI healthcare company leader, it seems like a great place to start would be to ask you exactly how artificial intelligence, or AI, is expected to impact healthcare.  

John Bertrand:  AI’s potential impact on healthcare stems from the recent advancements in data digitization and standardization. Look at the way electronic medical records (EMR) have progressed in the last 15 years. We went from data being very disparate, often recorded in multiple systems or on paper, to a much more centralized, curated, and managed pool of data. 

What excites me about Digital Diagnostics, and why I decided this is the place I absolutely had to be, is that we’re in this moment of time that’s starting to open up where the data has been aggregated and standardized. Additionally, AI science and understanding are finally hitting a point of maturation, where you can see the two intersecting and stripping a ton of costs out of the system. This, in turn, is expanding access for patients by offering specialized care in a more equitable and fair manner.  

For me, I look at the state of AI as if we’re in inning two of a baseball game. We finally have the data, AI has been largely demystified, and we are starting to see early applications gaining real traction in terms of use and utilization within the healthcare system. At the same time, however, there is still quite a bit of innovative work left for us to do.  

Digital Diagnostics: LumineticsCore™, Digital Diagnostics’ flagship product, is designed to seamlessly integrate into clinical workflow. Can you share some examples of AI applications in clinical workflows that you have observed throughout your career, and how have these applications evolved over the years? 

John Bertrand: As I mentioned, we’re starting to see the first phase of AI applications integrating into clinical workflow and having a real impact. When I made a career shift post-EMR phase, I worked for early-stage investors. I remember looking at different things in labs and research institutions around the world and acknowledging how incredible the technology was, but at the same time realizing it would be quite some time before it was used and adopted.  

Those examples are probably six or seven years old now that I think back, but we’re now at a point where there are some really good use cases. For example, there are two early companies that are really driving exciting clinical outcomes in pathology. While you might assume they are in just one or two early adopter locations, they are in a couple dozen facilities and still scaling.  

There has also been a great deal of advancement in the cardiac monitoring space, where AI is often used as an assistive tool to compress result turnaround time for patients. Again, that technology is operating at scale at a couple companies.  

At Digital Diagnostics, we’re at a point where our technology will be in thousands of individual facilities in the near term. You can really look around the industry at this point and find several companies that are scaling with AI in ways that go beyond mere add-ons, bells and whistles, or marketing tactics.  

One thing that I’ve been cautious of over the years is claims of groundbreaking new AI. Often, these turn out to be either not true AI or AI that fails to drive the type of value that this technology is capable of. Too often it’s just an incremental add-on used to increase product margins. In contrast, the companies I’ve mentioned are unlocking serious patient value, driving better clinical outcomes, removing costs from the system, and expanding access. That’s what I’m referencing when I talk about early adoption and the current role of AI in healthcare. 

Digital Diagnostics: When creating LumineticsCore™, company founder, Michael D. Abramoff, MD, PhD, knew patients, physicians, and the public might have some reservations about trusting a computer, rather than a doctor, to make a diagnosis. Based on your experience, how do people generally react to AI in healthcare – are they more trusting or skeptical? Do you have any suggestions for mitigating uncertainty surrounding healthcare AI?  

John Bertrand: In the early days, there was widespread skepticism. I’ve seen the skepticism firsthand at several companies during their initial product launches. When I came to Digital Diagnostics and started working with our founder, Michael D. Abramoff, MD, PhD, we saw the exact same skepticism. It’s not just an issue with AI, though. I would say it’s a common reaction to any new technology in healthcare. Generally, nobody wants to be disrupted.  

Healthcare is complicated and a high-risk environment. If you make a mistake, someone could die. It’s very serious. I recall a time early on in my career when I realized the gravity of my responsibilities. It dawned on me that if I didn’t design things correctly, leading to data being misplaced or overlooked, I could be responsible for harming somebody. It’s a serious deal, which is why it’s crucial to focus on a couple specific things.  

One focal point should be the rigorous validation of the product. Companies should be able to provide transparent documentation of this process and the outcomes. For us at Digital Diagnostics, that was our FDA De Novo clinical trial. It’s crucial to work within the system to validate the product, ensuring you have those very tangible, concrete proof points.  

The second area of focus should be the product’s impact on patients – not the technology behind the product. In the end, the technology isn’t what people care about. Companies must be ready to discuss whether the technology is driving improved patient outcomes, how they know it’s safe, if it fits within a standard of care that’s already validated as beneficial for the patient, etc.  

The idea of ‘first do no harm’ is paramount, followed closely by considering the new product’s impact on provider workflow since we all know healthcare professionals are overburdened. Additionally, there are ongoing labor issues in the healthcare sector that are further exacerbating their problems. While there are numerous challenges providers face, it’s necessary to succinctly explain how a newly developed technology integrates into their workflow and simplifies tasks for everyone involved.  

Those two things are key to garnering acceptance for new technologies in healthcare. On the other hand, I believe it’s ineffective to pitch a new technology by claiming it’s disruptive and will change healthcare, particularly if it’s a new technology that’s not widely understood. It’s just not a great way to talk or think about it.  

Digital Diagnostics: In 2018, LumineticsCore™ became the first FDA-cleared autonomous AI diagnostic system in healthcare. In recent years, several other AI healthcare companies have garnered recognition for their contributions to patient care. In your opinion, which companies stand out as major players in healthcare AI and why?  

John Bertrand: We must start by defining what we mean by major player because everybody that’s reading this probably has a different opinion. Many people see what companies such as Google, Amazon, Microsoft, and Apple are doing in healthcare and say those are the big players. Personally, I take into account who’s doing something that hits the checkboxes I mentioned earlier, such as driving value, improving patient outcomes, enhancing access, and promoting equity. 

The people that I consider major players are doing things that are very tangible right now. I think Digital Diagnostics is a great example of that. There are other companies that are checking those boxes as well. For instance, the companies Path.AI and PaigeAI are great examples. In my opinion, they’re driving real  value in the pathology space and hitting the checkboxes I mentioned before.  

iRhythm is another company that’s successfully scaled AI to add value for patients. Though I don’t have the specifics, I’d assume that at this stage, iRhythm is likely processing the data of millions of patients annually. When I’m making decisions in my own business, I always find myself thinking, “What approach did they take to solve this problem, before committing to a particular resolution.  

Digital Diagnostics: Digital Diagnostics is an AI healthcare company, and as such has always operated within the healthcare space. However, as AI technology becomes more popular, we’re seeing large tech companies not traditionally involved in patient care venturing into healthcare AI with varying degrees of success. In your opinion, what factors contribute to the difficulties faced by big tech companies entering the healthcare space?  

John Bertrand: I think a common mistake in healthcare AI is when these large tech companies begin the product design and development process with the mindset of gathering data and utilizing technology for the sole purpose of monetization. This approach is flawed because it doesn’t begin with a problem.   

The guiding principle here should be to first understand the problem that needs solving. The technology should be the tool used to solve the identified problem. It’s when the process is reversed that issues arise – when technology is in search of a problem. Similar to a hammer trying to find a nail, it’s not particularly productive.  

From what I’ve observed in the marketplace, this often leads to situations where companies try to justify the products they’ve created without any solid clinical validation. They’re unable to demonstrate how their products are driving an improvement in outcome, reducing costs, expanding access, or improving equity. If a product isn’t doing those four things, it’s not valuable in my opinion. Oftentimes, when companies start with the technology first and then search for a problem, they end up trying to justify the work they’ve already done, which I find unsettling.  

Another concern for me is when a healthcare AI company is comprised entirely of engineers and lacks clinicians. In developing AI solutions for healthcare, it’s important to consider the same factors a clinician would consider when making treatment decisions. The absence of a physician’s perspective can lead to products that may not effectively integrate within the existing healthcare system.  

Digital Diagnostics: Since its founding, Digital Diagnostics has worked from within the healthcare system to establish automated diagnosis as the new standard of care. In your view, are adoption and integration of AI technologies into the healthcare system possible if all stakeholders aren’t engaged? 

John Bertrand: I don’t believe adoption and integration of AI technologies into the healthcare system are possible if all stakeholders aren’t engaged. This isn’t a type of industry you can just ram a product through. It’s essential to have buy-in from everybody, and that’s a conscious decision we’ve made at Digital Diagnostics. We work from within the system.  

Every company I’ve mentioned throughout this discussion is the same way. They collaborate with regulators, engage with scientific thought leaders, and, if their product affects public health, they consult with experts in that field. This type of broad engagement is crucial. There’s a rationale behind this approach, namely, patient safety.  

If you want people to trust the technology or the product that you’ve built, it’s imperative to work within the system. Attempting to reproduce Uber’s strategy of disregarding others and forcefully entering the market just won’t work in healthcare. It violates too many concerns about safety and efficacy. 

Furthermore, healthcare is built on the “first do no harm” principle. And if a company isn’t collaborating with those who built, manage, and adjudicate the technology, it will be difficult to determine if that technology is truly doing no harm. When a company operates in this manner, it’s unlikely to be successful, in my opinion.  

Digital Diagnostics: Since you have been involved with Digital Diagnostics throughout the journey of introducing a novel AI technology to the market, what work would you say is left to do to drive adoption of AI in healthcare?  

John Bertrand: An ideal next step in driving adoption of AI in healthcare would be standardizing the process of bringing AI-based technologies to market, specifically creating well-defined guidelines for navigating the FDA’s submission process. The FDA is working on that. They’ve published many papers on the subject. Digital Diagnostics’ founder, Michael D. Abramoff, MD, PhD, has been involved in many of those papers, serving as a thought leader.  

It’s also crucial that we standardize the way reimbursement is calculated. The current systems and structures are built around the assumption of physicians being involved because that’s how it’s always worked. Once we know the framework with which AI safely functions and adds value, we can update the regulatory processes used to bring these products to market.  

When each stakeholder agency has been positively engaged, we find that they want to collaborate with innovative companies, which I found pleasantly surprising, as someone who hadn’t really done this type of work prior to joining Digital Diagnostics. It’s been interesting to see their commitment, mirroring ours, to doing the right thing. I see a shift towards standardization, and I truly believe that is the next critical step. We’ll get there, but I think it will be another year or two due to the complexity involved. There are many factors to consider and numerous stakeholders to involve to ensure it’s done properly.