Table of Contents
Bhaskar Sambasivan, CEO of Saama, sat down with Onyx for a feature-length interview.
Could we begin by exploring your background and how you came to be leading Saama?
I've been working at the intersection of healthcare and technology for almost 30 years across different types of companies—services companies, consulting firms, product companies—all within the healthcare and pharma space. Prior to Saama, I was the CEO of a product engineering company called CitiusTech. Before that, I was with a company called Eversana, deeply focused on pharma commercialization services.
I also spent almost 15 years at Cognizant, building up the pharma and life sciences business unit, and spent time with Siebel Systems and consulting firms like PriceWaterhouseCoopers.
I've known and admired Saama since 2015, back in my days at Cognizant when we had built a product for clinical trials called Smart Trials, and we competed head-to-head. Saama has come a long way since then, continuing to evolve their products and platforms to drive efficiencies and accelerate time within clinical trials by leveraging AI as the foundational component—that’s the big differentiator.
Saama's name really came to the forefront when their platform was instrumental in helping Pfizer save time during the COVID pandemic and get their vaccine out sooner. My passion lies in making a difference by leveraging technology. That's what really excites me, and that's why I'm here.
Could you give us an overview of what Saama offers and how the technology works?
Saama's mission is to accelerate the speed of drug development by integrating advanced, scalable, and proven AI models. Relying on accurate, high-quality AI models that can scale to hundreds of studies, thousands of patients in a phase three trial with hundreds of millions of data points is not an easy task.
That's a big focus area for Saama. When you think about the clinical trial workflow, these have been carried out by sponsors and CROs for years in traditional ways. But as pharma moves to more complex therapeutic areas—specialty diseases, rare diseases, more focus on oncology—the trials become more complex.
What we've done at Saama is create our own bespoke LLM (Large Language Model). We've further enhanced it with additional datasets and have our own version called Clinical-LM, which we use as the foundational layer that powers all our products.
It's a fraction of the cost because it's based on open source software, and is trained to address specific needs in clinical and biomedical domains. It has been proven with some of our products, and with pharma companies like Pfizer and others.
The product in our partnership with Pfizer is called Smart Data Quality. It essentially helps in end-to-end clinical data review, getting to a database lock sooner so data analysis can start.
The second product we have is called Patient Insights, which helps in data review from a safety standpoint—patients at risk during a clinical trial. What are our signals? How do we detect a safety or an adverse event sooner based on the data?
The third product is called Operational Insights. Think of it as a cockpit to tell you where you are in the trial, how sites are performing, how patients are being enrolled, so that study teams can always think about the next action or intervention.
Underlying all of this is our Data Hub product, which serves as the foundational clinical data store, which is vendor-agnostic and data-source agnostic. You can bring in EDC data, lab data, eCOA data, ePRO data, sensor data, biomarker data. We already have APIs with all the leading EDC vendors, CTMS vendors, and other technology vendors or data partners.
We use that as the foundational Clinical Data Hub where the data gets standardized and is ready for use by any of the three applications or any other tools or platforms that our clients can bring in and plug into the Data Hub.
So we need the Data Hub, the foundational Clinical-LM large language model that I just talked about, and that's how we go to market. That's our biggest differentiator. We'll continue to learn, continue to train, and get our models more powerful. Today, there are over 100-plus models in any single product that I talked about.
We are working on over 700 clinical trials across all the different phases, covering almost 115 therapeutic areas, including vaccines, rare diseases, and a big part of what we do is oncology as well.
So with the LLM you mentioned, can you give us an example of how the end user might interact with it?
Let’s take Smart Data Quality. The best use case there is typical data management. Clinical data management or data review happens as patients get enrolled in clinical trials—their data is entered by sites into the EDC. It’s very important in pharma to get to a database lock. Only after the database is locked do we have statisticians and SAS programmers looking at the data for analysis, integration of results, insights, and so on.
Traditionally, that involved an army of people: data managers, data reviewers who look at the data, look at discrepancies, look at outliers, raising queries, going back and forth - and that’s time-consuming. An inexperienced team who doesn't truly understand the data, and why it's entered in such a way, may be raising 40-50% non-critical queries, sometimes even more.
Our AI model acts on the data, and based on past learning and training, it points out discrepancies or outliers, recognizes which are important, and raises queries automatically. What typically happens in a manual review process, AI does automatically.
With AI, you always have to have the human in the loop to validate and verify—not only from a regulatory standpoint but also to ensure that what the study teams are doing is absolutely accurate and high quality. But it significantly saves time as opposed to going back and forth through emails or phone calls.
Saving time, saving money, reducing site burden, and improving quality. That's how our Smart Data Quality product leverages AI, and that's how we continue to work with Pfizer and other large pharma customers.
Do you have a data moat? Is there a bank of proprietary data that you're starting to build, and what's the main area of focus with that?
No, we don't own any data. It's all our customers' data—pharma's data. The data is owned by the pharma companies or the patients themselves when it's patient data. We process it, orchestrate it, standardize it, clean it, and make it ready for insights and analytics.
Looking forward to 2025, what are your main goals? Where exactly would you like to lead Saama?
Before I focus on Saama, it's important to spend a minute or two on how I'm thinking about the market conditions. It's unfortunate, but multiple forces are coming together at the same time. There are the macroeconomic conditions. Also, policy reforms like the IRA (Inflation Reduction Act) are adding significant pressure. Almost every pharma company has embarked on some kind of cost-saving initiative as a strategic priority for the CEO.
In the past 10 years, R&D returns have significantly declined while the cost to bring an asset to market has significantly increased. The cost of conducting clinical research continues to increase because it's very slow and resource-intensive.
But the most important shift we are all witnessing at this point is the advancement of GenAI. Organizations will embrace the shift or fall by the wayside.
That's what we are seeing right now. Every organization is rethinking its business and operating model to avoid being overtaken by competitors. Saama is at the point where all these dynamic forces are coming together.
Drug development is all about doing more with less. It's about driving efficiencies, being innovative, so you don't get overtaken by the competition. Our goal is to continue building high-quality AI-enabled products and platforms that will look at every step, every process, every function, focusing on maintaining or improving quality in less time and at lower cost.
And what other trends do you see in the wider industry?
There's a lot of scientific advances that pharma is thinking about, especially a focus on personalized precision medicine, leveraging omics data—proteomics, genomics, and metabolomics.
Even though R&D spend is under pressure and pharma wants to do more with less, there's increased activity in mergers and acquisitions with a lot of venture capital funding going into biotechs and early-stage ventures. So definitely, pharma is poised for innovation. I'm truly excited and looking forward to it.
