Predictions Series 2022: AiThority Interview with Dr. Jack Zeineh, Co-Founder and CTO at PreciseDx

Hi, Jack. Welcome to our Interview Series. Please tell us a little bit about your journey and what inspired you to co-start PreciseDX?

I have been interested in computers and their capability to solve problems since high school when I started writing software for electrophoresis analysis using the Apple II I had been using for gaming. I’ve been involved with digital pathology since finishing medical school. After growing a company focused on digitizing pathology slides, which was ultimately acquired by Carl Zeiss, the utilization and complete harvesting of the information contained in those images became a clear next step.

I believe that Artificial Intelligence is the key to unlocking the potential of this vast trove of data contained in Pathology slides, PreciseDx, was borne out of this belief.

How do you leverage AI at your organization?

AI is the cornerstone of our technology. It allows us to fundamentally recast how image are analyzed in pathology using methods that are more consistent and accurate than humans interpretation alone. This then enables us to address problems that previously were largely intractable in terms of creating commercial solutions to clinical problems.

Recommended: Predictions Series 2022: AiThority Interview with Dr. Arnaud Rosier, CEO & Founder at Implicity

Could you please help us define “precision medicine” and how this science has evolved in the last 2-3 years?

Precision medicine can broadly be seen as utilizing to the fullest available patient specific data (as opposed to perhaps broad heuristic rules or population data) in order to make the best possible decisions for the patient. What we have been developing at PreciseDx is an example of this, we apply AI on pathology slides and provide a detailed set of quantitative results that are specific and unique for the individual. This facilitates the precision of the applied therapies used for that patient. A further example, within cancer, targeted therapies that are selected for a patient based upon the presence of activated genes in an individual patient’s tissue. For example, the use of immune checkpoint inhibitors such as Merck’s’ Keytruda have been used in patients that are over-expressing the PDL1 gene.

Why use AI for precision medicine?

AI is well formulated to augment and scale human capability in medicine by providing increased accuracy and repeatability of repetitive tasks such as cell counting, measurement, etc. This frees up available human bandwidth to do higher level review and clinical decision making.

Please tell us more about your AI and Deep Learning projects at PreciseDX. How do these capabilities align to solve the modern-day healthcare problems?

We’ve designed a platform we call a Morphology Feature Array™ that analyzes millions of data points from a patient’s tissue slide to produce a set of standardized features that can accurately, and reproducibility produce objective patient specific information that can be used by the clinical care team to create more appropriate management decisions for that patient. The other factor is we image and process the entire slide looking at every cell in the image, whereas conventional analysis looks at a sampling portion of the slide.

Recommended: Predictions Series 2022: AiThority Interview with Anoop Ramachandran, Chief Technology Officer at Preciso

How much has AI in healthcare evolved in the last 2-3 years?

AI has matured both technically and direct clinical utility. While the early days of AI in healthcare were hallmarked by academic papers showing promise, recently this promise has started to translate into real world, commercial applications that can be used in the clinic.

How can AI complement clinical care teams?

Furthermore, we see in the literature, and in several commercial organizations, the evolving embodiments, and applications of AI in pathology. These range from using AI to reproduce what a pathologist can already do (e.g., breast cancer grading and recreating the Nottingham Grading score), to improving the QC function in pathology and reducing false negative rates in cancer detection, to performing arduous tasks that are hard for the pathologist to do (such as accurate enumeration of mitotically active cells across the entire histopathology slide) to new capabilities that expand beyond traditional pathology determinations, such as providing better risk stratification and prognostic risk assessments than traditional histopathology.

Your take on new-age generative AI models and how these could improve outcomes for telehealth and diagnostics?

Generative models are one part of the bag of tricks available in order to solve a problem. They are often an important part but not a sole source silver bullet.

Recommended: Predictions Series 2022: AiThority Interview with David Low, CMO at Talkwalker

What is your take on the buzzwords swirling around AI-driven oncology and pathology?

This is natural for a technology still working its way through the system and simplified lexicon helps the propagation of ideas even if they are overused sometimes.

Any advice to every healthcare business leader on how to use AI and ML technology for precision medicine outcomes?

AI and ML technology are critical levers to improve quality and lower cost and ultimately benefit the patient. Solutions that are not just technically impressive but well connected to clinical workflows and disease models offer the best promise of successful integration.

Thank you, Jack! That was fun and we hope to see you back on AiThority.com soon.