This is Conversations with Digital Diagnostics, a series where we delve into insightful discussions with company thought leaders. This time, we are having a conversation with Juli Goldstein, Vice President of Government Affairs and Market Access at Digital Diagnostics.
Juli has a background in public health. She began her career in international health and has her credential as a Certified Professional Biller (CPB). Following her academic career, she transitioned to working internationally and, later, in the United States with some of the largest payers in healthcare. In the mid-1990s and early 2000s, she worked at a large national health plan in the Bay Area that also included hospital systems, provider groups, and medical centers across the country. From there, she became involved in the medical device industry, as well as federal policy and advocacy groups. In 2020, Juli joined the Digital Diagnostics team.
Digital Diagnostics: With your extensive background in healthcare and as the current Vice President of Government Affairs and Market Access at Digital Diagnostics, could you share your perspective on the challenges of managing diabetes, especially when it comes to following medical guidelines such as annual eye exams?
Juli Goldstein: Let me begin by sharing that diabetes is near and dear to my heart, as several close family members live with the condition. My understanding of diabetes comes from both personal and policy-related experiences.
Clinical guidelines suggest that those with diabetes should monitor their hemoglobin A1c (HbA1c) levels, undergo foot exams, and check their cardiovascular health every three months. However, daily realities often take precedence over the condition. So, adding another appointment, such as an annual eye exam, can seem daunting and is frequently overlooked. As you can imagine, managing diabetes is not the primary focus for most; living life is.
The recommendation to get an annual eye exam is grounded in prevention and supported by substantial evidence. However, because diabetic retinopathy often has no symptoms, it’s really common for those living with diabetes to think that they don’t need an eye exam. This perspective helps explain a significant problem here in the United States, where we have upwards of 30 million people living with diabetes, yet many do not get annual eye exams. In fact, research shows that, in some cases, only 10 to 20 percent of Medicare beneficiaries receive their annual eye exam for diabetes.
Digital Diagnostics: How did the integration of LumineticsCore™, Digital Diagnostics’ flagship product designed to diagnose diabetic retinopathy at the point-of-care, address the unmet needs of patients and clinics? Were physicians receptive to this technology, and how did it change their approach to diabetes care?
Juli Goldstein: Like any innovation, it all began with addressing the unmet need of patients and clinics. From the perspective we discussed earlier – of having one more task to complete or exam to undergo – you begin to understand the challenges faced by primary care physicians and family medicine practitioners who care for these patients.
There is a quality measure for comprehensive diabetes care that, in addition to HbA1c exams, includes annual eye exams for people with diabetes. Meeting quality measures for diabetes care can be difficult for primary care providers, who already face numerous demands in their clinics. This situation highlighted a clear unmet clinical and patient need. Our founder, Michael D. Abramoff, MD, PhD, who is both a retina specialist and a computer scientist, recognized this gap. From his personal experiences in the clinic, he often encountered patients with diabetic retinopathy so advanced that they had irreversible vision loss.
Understanding that more than half of people with diabetes will develop diabetic retinopathy, which can lead to vision loss, underscores the importance of early intervention.1 This clear unmet need provided a compelling reason to develop a fully autonomous diabetic retinopathy exam. Dr. Abramoff was particularly driven by this urgency and initially brought the concept to the FDA. It took him over a decade to achieve clearance, partly because the idea involved a machine performing tasks traditionally done by physicians.
This innovation was not just about utilizing artificial intelligence for clinical decision support, though. It was that the artificial intelligence would take the role of providing the diagnosis. Interestingly, the initial reaction we often observe, whether from payers, policy makers, stakeholders, or physician groups, involves concerns about how AI will perform tasks traditionally done by doctors.
But when you recognize that this technology allows frontline care providers to immediately show patients the impact of diabetes at the point-of-care, adjust medications, and make a referral to eye care specialists, including optometrists, you start to understand that it’s more than just a machine doing a diagnostic exam. It’s about enabling providers to work at the top of their license and facilitate primary and frontline care to effectively do the jobs that they need to do. This process involved a shift in how we think about and bring innovation to market from FDA clearance to securing payment.
Digital Diagnostics: In 2018, LumineticsCore™ became the first FDA-cleared autonomous AI diagnostic system in healthcare. How critical was the FDA’s involvement in the years leading up to this clearance? And what role did other stakeholders play during this period?
Juli Goldstein: In healthcare, stakeholder engagement is critical and should always be a top priority. In terms of the FDA itself, Dr. Abramoff began engaging with the FDA ten years prior to obtaining De Novo authorization, or clearance.
One of his initial steps was to engage ethicists, given the complexities of employing artificial intelligence and machine learning where machines perform tasks traditionally performed by physicians. This proactive approach was vital in understanding that regulators would require substantial support to navigate these innovations.
That ethical framework is foundational to everything we do, with ethicists playing a crucial role as key stakeholders. Additionally, engaging FDA administrators was essential, especially regarding discussions about De Novo authorization, as this was the very first clearance of a fully autonomous artificial intelligence system in healthcare.
Extensive collaborations with ethicists and the FDA as well as policy makers and patient and physician advocacy groups were vital. We spent considerable time determining how to conduct a clinical trial, what endpoints to consider, and how to prove the system’s safety and effectiveness. We also had to establish reference standards for measuring effectiveness and define what it means for artificial intelligence to operate fully autonomously without the need for physician oversight.
Interestingly, stakeholder engagement became so critical that it defined our pivotal clinical trial and influenced the FDA’s decision to grant breakthrough status to this device. This focus on collaboration ultimately led to the first FDA clearance of a fully autonomous artificial intelligence in any field of medicine.
Digital Diagnostics: Can you delve deeper into Digital Diagnostics’ journey from FDA clearance to achieving widespread adoption, particularly on the challenges and milestones related to attaining payment and reimbursement for a novel AI technology and why reimbursement and payment are so crucial for the adoption of new or novel technologies into the healthcare ecosystem?
Juli Goldstein: The journey to FDA clearance took about ten years. After receiving FDA clearance, we thought, “Great, we’re clear to go to market. We’re going to make our shareholders and providers very happy, and we’re going to solve the problems of healthcare.”
However, we quickly realized the critical importance of payment and reimbursement. Without a designated CPT code, established price, and coverage, it is challenging to drive provider or, most importantly, patient access. In the United States, these factors are pivotal in determining reimbursement strategies. So, the process for FDA clearance, while paramount and a decade long, began the next stage in the journey, which was attaining national reimbursement and coverage for our fully autonomous artificial intelligence system.
The stakeholders involved shifted over time. Ethicists and the FDA remained, but we also began to focus on The Centers for Medicare and Medicaid Services (CMS), provider groups, advocacy groups, specialist organizations, frontline providers, and patients. We obtained FDA clearance in 2018, and by 2021 CMS established a category one CPT code for the first time. A category one CPT code is important because it is necessary for providers to use a technology. Without a code, it is very difficult for technology to be widely adopted or to be able to deliver benefits to patients.
As a company, we noted that FDA clearance marked an increase in utilization and access to this service. However, when the code and national payment were implemented in the subsequent years, adoption surged. Utilization and access to this technology across the country – in both rural and urban areas – began to shift, and that is because of reimbursement.
We often get asked, “Why should we care about reimbursement? If there’s an innovative product that fundamentally changes healthcare, why focus on reimbursement or payment?”
The answer’s simple. Providers are busy and must be compensated for the services they deliver. In the United States, we operate under a fee-for-service foundational model, where providers are paid based not only on their productivity but also on the quality of their service. I frequently encounter questions like, “Why do we have to focus on fee-for-service? Why not just concentrate on quality?” However, quality is inherently linked to the fee-for-service payment model. Without this model, it becomes very challenging to access healthcare services throughout the United States.
Digital Diagnostics: Given your extensive background in healthcare and your work with Digital Diagnostics, can you provide some insight into the public and professional reactions about AI replacing jobs and its potential impact on the healthcare industry?
Juli Goldstein: Let’s look at the adverse side for a minute. There’s an editorial, where Dr. Abramoff was nicknamed “the Retinator,” which stemmed from skepticism about the role of machines in healthcare. At the time, there was this whole pushback with critics’ questioning, “Will machines take the jobs of providers?”
Concerns about the impact of AI on employment remain a relevant topic today. It’s still debated whether AI will replace writers, waiters, and even drivers of taxis and Ubers. Similarly, AI’s impact on healthcare continues to be debated. The earlier publication of these concerns has been beneficial, as it underscored the need to educate policy makers, advocacy organizations, and thought leaders about AI’s potential to improve access to care, reduce healthcare costs, and close the gap in health equity.
Digital Diagnostics: Could you discuss the implications of the 2020 updates to standards of care and quality metrics, especially considering that at the time, LumineticsCore™ was the first and only FDA-authorized autonomous AI diagnostic system for detecting diabetic retinopathy? How have these changes impacted health plan policies and the accessibility of healthcare services?
Juli Goldstein: Let’s first discuss the standards of care. In 2020, significant policy updates were made in some of the largest health plans in the United States. These health plans acknowledged the potential of services driven by autonomous AI to fundamentally change healthcare, leading them to update their health technology assessments to specifically include artificial intelligence for diabetic retinopathy eye exams as recommended and medically necessary. This update marked a significant shift in the standards of care.
At the same time, the quality metrics underwent changes. Previously, quality metrics only acknowledged feedback from an eye care specialist to confirm a patient had received their required eye exam for diabetes. However, the quality metrics were updated to reflect the growing acceptance of healthcare AI, particularly the evidence showed that in some of the most inaccessible areas of the United States, LumineticsCore was instrumental in improving access and advancing health equity. The exams performed by LumineticsCore, when implemented in frontline care settings, significantly improved service delivery.
Digital Diagnostics: Digital Diagnostics’ tagline is “AI the Right Way.” What exactly does this mean and how has Digital Diagnostics’ commitment to AI the Right Way shaped its approach to developing and implementing AI technologies in healthcare?
Juli Goldstein: Digital Diagnostics uses the tagline AI the Right Way, reflecting not only the development of AI and machine learning technologies, but also their underlying fundamental dependence on ethical development.
From the very beginning, we committed to ethical development, aiming to work within the healthcare system, not disrupt it. So, we would approach it the same way if we were to do it again. Engaging with stakeholders – from policy makers to regulators, advocacy and patient groups, and physician and provider organizations – was critical. Addressing the unmet needs of patients and the benefits to them was the right place to begin. We wouldn’t change our approach to navigating through the long, arduous process required to gain FDA clearance.
Setting up an ethical framework from the start has allowed for widespread adoption later on. For other innovators out there, it might seem tempting to skip some of these long-term steps, but it’s these steps – collaborating with the FDA and incorporating insights from ethicists – that have fundamentally enabled the widespread adoption that we see today. We would never have been able to attain this level of adoption without the early buy-in from these stakeholders.
References: