Table of Contents
What experiences and turning points led you to found Prognos Health?
I’ve basically been a health-tech entrepreneur throughout my career. I’ve built and sold several businesses, all of which were in health tech. Before starting Prognos, I started companies like Medsite, which we sold to WebMD. Jason was the co-founder. We were talking about how to have impact, at scale, for patients, and we came up with the idea for Prognos Health, because Jason’s a physician.
So, he was thinking about it from the perspective of data points that have the biggest impact on patients’ lives. And he said that about 70% of the decisions that he makes are based on laboratory and diagnostic results. I had already built a couple of companies in the space. From an industry perspective, it seemed like a great opportunity, because lab data was just being underutilized.
With precision medicine and biomarkers becoming so important, it seemed like the industry could really benefit from having access to lab data at scale. So, we came together and started with the idea of having a big impact upon patients by leveraging a really important dataset that really wasn’t being utilized.
If you were to describe an overview of your platform, how does it operate and what AI capabilities differentiate you?
We’re a healthcare data marketplace and we partner with labs and diagnostics companies. We’ve built a platform that leverages AI - machine learning, NLP - to take data that comes unstructured and in different formats. Within hours, using software, we can turn that into a standardized, curated dataset, that our customers can utilize within minutes.
There are really four things we do - our PACT promise. We’re the most precise - the ‘P’ - when it comes to lab data because we have the same underlying information that physicians look at. Everything we deliver to pharma is actionable - that’s the ‘A’. It’s comprehensive - our ‘C’ - and we work with over 20 lab partners across national testing, specialized oncology testing, and genetic testing for rare disease. And the ‘T’ stands for timely - because we’re leveraging AI and software; from the time that the data hits our system, we can turn it around and have it delivered to clients in as little as four hours.
Do you almost see yourself as an external auditor of Phase IV for the FDA?
I think the role we play, today, is lab and diagnostics data. We don’t get hospital provider or EHR data - that’s available through others. But someone could link what we have with that and do Phase IV-type monitoring. It’s not something we do today, but we handle the lab-data portion of that.
Where is your data mainly coming from? U.S.? Are you looking abroad?
We’re 100% U.S.-focused, right now.
Could you give an example of a potential client, the type of data they’d want, and what they might do with it?
Yeah, the data that’s coming to us - if you think about labs, they’re very test-focused. But pharma really thinks about their business at a therapy and disease level. So, all this data is coming in unstructured and test-level, and we’ve basically built profiles by disease areas.
We know all the relevant tests that matter for diagnosing or monitoring patients with that disease. So we’re building therapy-area-specific datasets that clients can subscribe to.
Examples could be a rare cancer, and looking at a certain genetic mutation, like a BRCA mutation. The data that’s coming in every day is across all these different tests. We create a dataset specifically for breast cancer, and it’ll have the biomarkers or genetic mutations - positive or negative - which is updated in real time.
It’s anonymized at the patient level but can be identified at the provider level. So, all their outreach - medical science liaisons, sales teams, and so on - can go talk to physicians about the patients, testing, and how the right therapy can help. Or they can activate their marketing outreach using the same information.
Are there particular therapeutic areas or disease segments you’re especially focused on?
Yeah, about 80% of our customers are in oncology and rare diseases. So, from areas where lab results and biomarkers really help with diagnosis and treatment decisions - that’s where our focus is.
The remaining 20% are in broad conditions - diabetes, cardiovascular risk and asthma, for example. Eventually, I think every disease will have biomarkers, but, today, it’s mostly oncology and rare diseases.
Any examples of real-world patients Prognos have helped?
Yeah, we track this very closely. Our BHAG is to deliver 20 billion health insights by 2050. We’ve delivered about 1.7 billion, so far. Each insight is information that helps a physician make a better decision or helps a patient get the right diagnosis or therapy.
In rare disease, hereditary angioedema is a great example. Sometimes, it takes a decade to diagnose. Using our data, we can help inform the right diagnosis within hours. Think about the impact for a patient who’s been misdiagnosed for years.
In oncology, there are patients with specific mutations and targeted therapies that exist, but physicians may not know. They start with chemo because that’s the default. But, with our data, they can find the right mutation and match the therapy much earlier.
My mind goes to JFK’s granddaughter - the rare blood cancer story in the news.
Great example. Exactly the kind of thing our lab, biomarker, and genetic data can help with.
And that 1.7 billion - wasn’t it like 0.7 billion more than a few months ago?
It’s taken us six or seven years to reach this point, but yeah, it’s pretty incredible. Furthermore, we can see a path to 20 billion. It’ll take a while, but we’re committed!
What regulatory hurdles do you have to deal with?
One of our biggest concerns is to make sure the data we’re handling is treated with the highest privacy standards. We work with third parties - Datavant is a partner - and we use their software to tokenize all the data. Everything’s anonymized and tokenized.
We also work with other third parties to stay HIPAA-compliant and make sure nothing can be re-identified. We’re SOC 2, HITRUST - SOC 2 certified as well. We adhere to all the industry standards that you’d expect for healthcare data.
What exactly is your business model, and how do you balance the interests of payers, biopharma, and diagnostics clients?
We operate between healthcare providers, those who generate the data, and the pharma clients who subscribe to the data. It’s a marketplace model. Partners place their data on our platform. We handle everything else - standardization, enrichment, curation - and organize all of it by therapy area.
Pharma clients subscribe. It’s roughly a 70/30 split: data providers get 70%; we take a 30% marketplace fee.
When tracking impact or ROI for clients, what metrics matter most?
Most of the ROI tracking is internal to the clients. But we make sure they’re getting value. We stay engaged well beyond delivery, and we’re always getting feedback on what features of the data matter most.
Our net retention rate is over 100%, so we know the value is there. Some have shared ROI of five-to-one or higher!
Are you pursuing strategic partnerships or alliances?
Absolutely, I’d say there are three types.
The first are companies that have the data, especially oncology and rare disease labs. We have critical mass, but we’re always expanding.
Secondly are infrastructure companies who handle healthcare data. Datavant is one. As we use their software, our data is interoperable with everyone in their ecosystem: claims, prescription, EHR data. That’s really valuable.
The third type of company we work with allows for our data to be accessed - cloud compute, so AWS, Google, Snowflake. Those are the ways that customers connect with the data.
As AI models proliferate and become easier to build, what’s your moat?
AI starts with the best quality of data that you can get. That’s where we’ve invested for the past seven or eight years: taking raw unstructured lab data and making it structured, standardized, and clean.
For AI models or agents, you need reliable data. Nobody comes close to us on lab and diagnostics data. Clients tell us this all the time, having worked with others, and even directly with labs. Once they work with us, they see the quality. Sometimes, they’ll literally ask if they can buy a lab’s data through us. That happens a lot.
Even when they contract directly with labs, they’ll ask us to integrate it all and give them one standardized deliverable. That shows the value, and the differentiation, that we offer.
What advice would you give to a new CEO trying to make a name in healthcare analytics?
It all starts with the data. Make the investments, make sure the data’s high quality and fit for purpose. Think about the end use case and build datasets and applications that solve a real problem people will pay for.
Last question - what’s it like running a company with your cousin?
He’s like my brother. There’s implicit trust. We’ve known each other since we were kids. He understands the clinical side, I understand the business side. It’s a great partnership.
What questions should we ask Sundeep next? Let us know in the comments.
