What made you decide to come to VIDA? What excites you?

I’ve spent the majority of my professional career providing tools to help radiologists deliver the best possible care for patients. VIDA is exciting because the company is applying intelligence to lung and respiratory care in a way that is truly transformational. This inclusion can have a profound impact on the management of diseases that continue to grow in prevalence, to the healthcare system burdened by a growing cost curve and most importantly, to patient quality of life. With over 200 journal papers published and over 30 established biomarkers, VIDA is in a great position to enhance the care for many patients across its 450+ sites that are using the company’s lung intelligence solutions today.

Tell me about your history in imaging and AI

I started out in bone densitometry for diagnosis of osteoporosis and from there moved on to work in the advanced visualization space. In my time at Vital Images, we partnered to sell Computer Aided Detection (“CAD”) products for lung, breast, liver and colon. CAD laid the groundwork for adoption of AI in interpreting object (visual) data, but its dependence on a rules-based coding made the framework too rigid for broad scaling and extremely challenging to improve. CAD also could often create delays in interpretation of data and have many false positives, both which became a source of frustration for clinicians. With the recent advancements in machine-learning and deep-learning, algorithms have the capability of greatly increasing accuracy while decreasing processing time. These advancements were why I joined Aidoc, which demonstrated radiologic value in acute, time-sensitive incidents across the body. They are the same drivers that have now brought me to VIDA, where I see additional opportunities for both time-sensitive and longitudinal value with specific focus on and expertise around a single anatomical system.

Why does AI matter in radiology?

I firmly believe that radiology is the most data rich specialty in healthcare. In addition to claims, demographic and CPT data, each image alone contains a figurative ocean of content. It’s been said that the naked eye is capable of deciphering fewer than 50 shades of gray, whereas a traditional CT image has more than 2001 levels of gray. An algorithm that can discern visual and quantitative elements with a scientifically meaningful degree of precision can deliver transformational impact for radiologists, helping them to deliver differentiating value to the physicians they’re serving. There is also significant efficiency gain when this technology is embedded into a clinical workflow seamlessly. In the case of VIDA specifically, we’ve found that clinicians average a time savings of 35% reading chest CT lung stacks, which are the most complex and time consuming of all radiology exams[1].

What about other physicians who apply radiology insights to patient treatment decisions?

The future for AI in clinical practice is unbelievably vast. We already see staggering amounts of funding in healthcare tech, yet many of these have siloed outcomes. AI – and in particular, the machine and deep-learning sub-components of AI – can help to accelerate and streamline the flow of data across systems. The result will be a huge relief to physicians who find themselves overburdened by administrative tasks. It will also create a meaningful shift in care coordination around the patient. Enhanced precision in radiology reporting not only provides front-line physicians with actionable insights to drive treatment decisions, but also helps to facilitate clear, informed conversations between patient and provider.

How do you think VIDA is different from other AI providers?

VIDA is different because it’s more than an “AI provider.” VIDA is a clinical company, with over a decade of focus in a very specific and very important field. This focus has resulted in a vast library of images, significant depth to product toolsets, over 200 journal papers and identification of over 30 biomarkers through that research. Where many tech-first companies develop an AI platform and look for a matching clinical application, VIDA began with deep clinical understanding and designs sophisticated algorithms best suited to building intelligence in this domain.

Why is this important?

Because modernization and transformation of care must start with a profound understanding of the clinical condition. As an example, 81% of Chronic Obstructive Pulmonary Disease (COPD) patients in this country are diagnosed at GOLD Stage II+[2], by which point symptom management becomes more challenging and costly. COPD currently affects 16 million[3] Americans, is growing in prevalence and is the 5th leading cause of death by disease in the US[4], recently eclipsed by COVID-19. Interstitial lung disease (ILD) has a misdiagnosis rate of 55%[5], with 79% of those patients having consulted at least three physicians prior to diagnosis⁴. Finally, when either of these conditions is comorbid with cancerous lung nodules or respiratory impacts of a disease like COVID, there are additional treatment and disease progression implications that won’t be apparent if the conditions are missed.

Imagine the impact at a patient level, for the family of the patient, for the health system bearing the cost of treatment for the patient; simply from increased accuracy in detection and diagnosis. Having a basis in clinical understanding of lung and respiratory disease places VIDA in a leadership position in providing automated quantitative data for both radiologists and pulmonologists. With the coronavirus pandemic, we are also now learning that COVID-19 may have severe lung impact in the near and long term. COVID-19 is a serious illness that VIDA is assisting healthcare providers with today. At a minimum, the tools we deploy to understand where disease resides and how it’s progressing should be fundamentally grounded in the organ or system of impact to ensure in that in a new disease we have a deep understanding of the context.

How should clinical buyers be evaluating different AI solutions?

The starting point is determining the accuracy of a given algorithm on their data. Next, IT administrators and physicians will want to think about ease of integration into existing systems and clinical workflow. How will they ensure clinical staff are interacting with the tools? How they can assess overall outcomes at the patient and disease management level, and efficiency gains that drive clinician satisfaction at work? Most importantly, they will need to consider the overall impact it will have to them as a provider. Determining key performance indicators ahead of any evaluation is of paramount importance so the solution’s value is undeniable and transparent.

Anything else we should know about you?

I couldn’t be more excited about my decision to join VIDA. AI is already transforming the delivery of healthcare and VIDA has been a clinical and technology trailblazer in the application of deep learning to modernizing lung and respiratory care. I believe in the transformational impact we can make for physicians who treat these illnesses and for patients who suffer from them. I know we are creating value in a part of healthcare where it’s long overdue, and I am proud to be part of the team leading this important work.

Sources

[1] Forsberg, Daniel, et al. “Radiologists’ Variation of Time to Read Across Different Procedure Types.” Journal of Digital Imaging, vol. 30, no. 1, Feb. 2017, pp. 86–94. DOI.org (Crossref), doi:10.1007/s10278-016-9911-z.

[2] Mapel DW, Dalal AA, Blanchette CM, Petersen H, Ferguson GT. Severity of COPD at initial spirometry-confirmed diagnosis: data from medical charts and administrative claims. Int J Chron Obstruct Pulmon Dis. 2011;6:573-581. doi:10.2147/COPD.S16975

[3] National Heart, Lung, and Blood Institute; https://www.nhlbi.nih.gov/health-topics/education-and-awareness/COPD-national-action-plan Accessed Sept 2020

[4] American Lung Association; https://www.lung.org/lung-health-diseases/lung-disease-lookup/copd/learn-about-copd#:~:text=COPD%20is%20the%20third%20leading,term%20disability%20and%20early%20death. Accessed Sept 2020

[5] Cosgrove GP, Bianchi P, Danese S, Lederer DJ. Barriers to timely diagnosis of interstitial lung disease in the real world: the INTENSITY survey. BMC Pulm Med. 2018;18(1):9. Published 2018 Jan 17. doi:10.1186/s12890-017-0560-x