On the heels of $111M fundraise, Zephyr AI taps biopharma exec Rachael Brake as it expands precision medicine efforts

Three-year-old AI precision medicine company Zephyr AI brought on board a scientist and executive with deep biotech and pharma experience to lead its strategic and commercial efforts.

Rachael Brake, Ph.D., is joining Zephyr AI as its new chief scientific officer after two years as CSO at Corbus Pharmaceuticals, where she built and led a novel research and development team focused on precision medicine and immuno-oncology. Brake also spent 11 years at Takeda Oncology, where she most recently held the head of U.S. medical affairs post for its oncology unit. She also worked in leading roles at Amgen.

The C-suite hire comes on the heels of the company's $111 million series A funding round, backed by Eli Lilly along with Jeff Skoll, eBay's first president, and EPIQ Capital Group.

Zephyr AI's tech combines data from multiple sources then uses machine learning and artificial intelligence to develop predictions about outcomes. The company says it's developing "fast and explainable artificial intelligence solutions" for precision medicine. Brake's drug development expertise ties in with Zephyr AI's strategic focus on curating large datasets combined with machine learning algorithms to build products and tools in the areas of oncology and cardiometabolic disease.

The U.S. has the highest rate of avoidable cancer and cardiometabolic-related deaths among any high-income country, noted Grant Verstandig, Zephyr AI’s co-founder and executive chairman, earlier this month when the series A round was announced.

Zephyr AI wants to use AI to extract novel insights to better define patient stratification and response predictions as well as improve federation of real-world data. Armed with the fresh funding, the company wants to accelerate its work to "democratize" precision medicine, enhancing both the speed and success of clinical trials, Verstandig said.

As a 20-year drug developer, Brake has spent most of her career, to date, understanding the molecular underpinnings of disease. "I've spent my career transitioning from basic discovery research into clinical development and then supporting scientific messaging on the commercial side of an organization," Brake said in an exclusive interview with Fierce Healthcare

She sees the promising potential of precision medicine to improve treatments for patients and the role that technologies like AI can play in advancing this work.

"Where we get it right, when you can deliver a precision therapy to a patient and you can marry that medicine to their disease, the outcomes are always better. I think we realize that it's difficult to do that effectively across multiple different tumor types. The most advanced place, obviously, is probably in lung cancer, but most of what drives that precision approach is really in the available therapies. And in the absence of the available therapies, then the premise of precision medicine, it's hard to execute against, and there's a ton of resistance," Brake said.

"Where AI and machine learning can assist us in that process is to offer a better opportunity to match and marry the underlying patient-specific information about their disease to a potential therapy. It also gives us the opportunity to traverse the normal boundaries of histology that we know, as scientists, that they're insufficient, but that's still the predominant paradigm for drug development. Bringing machine learning and AI into that context means that we can learn more about the operational networks that drive drug response and we can think about development paradigms that traverse that boundary of normal histology."

Having spent decades working in big pharma and biotech spanning roles in research, development and commercial, Brake comes to Zephyr with a unique breadth of experience, according to Jeff Sherman, Zephyr AI co-founder, interim CEO, and chief technology officer.

"She is a renowned scientist with an extensive background in oncology and precision medicine and a keen strategist skilled at developing and operationalizing cross-functional teams. Rachael’s mission-driven desire to simplify the complexities of personalized medicine is perfectly aligned with Zephyr’s commitment to address patients’ unmet needs and optimize outcomes.”

As a scientist, Brake said she is fascinated by the communication networks that are co-opted by cancer and how quickly they can adapt to outsmart current therapeutic strategies. "As a strategist, I am equally intrigued by the potential of AI to quickly evolve our approaches to treatment," she added.

"My premise in joining [Zephyr AI] is that I am a deep believer in precision medicine," Brake said. "Bringing AI and machine learning to that therapeutic context and background and expertise that I have, I see two opportunities. One is that I understand from a drug development standpoint, where the problems are, where the challenges are for drug developers, for physicians, for institutions, as well as diagnostic organizations, I've worked with all of the above. I can also think about where are the use cases in the real-world setting where we know a drug, it has drug-like properties, but it's underperforming relative to the promise and opportunity. And, there's biology as to why that is. Marrying these two things together, drug development and experience and expertise and this next generation of precision medicine, it's a natural marriage."

She added, "We've got to do a better job of crossing these disciplines together in such a way that we're more effectively using the tech in a way that is advantageous for drug development. "

Zephyr AI was founded in 2021 by investment and incubation firm Red Cell Partners, led by Grant Verstandig

Verstandig served as UnitedHealth’s chief digital officer until returning to venture investing at the end of 2021.

Zephyr AI has built a massive healthcare dataset through partnerships and proprietary data sources. The company's dataset of de-identified patient data now represents roughly 68 million patient lives, according to Brake. The company has built its AI and machine learning expertise to develop AI algorithms specifically designed to handle the complexities of real-world patient data. The company's algorithms then generate interpretable drug response predictions from that data, which are then validated with proven models.

A key capability is drug sensitivity prediction technology. "We can understand what might be the underpinnings of what drives sensitivity toward a medicine. Correlated to that, what we can also do is use very basic genomic information and clinical information and marry that from a patient in the real world setting, so patients in the wild, and understand what is going on with their cancer and understand what that predictive drug sensitivity may be. When we marry our technology to then real-world data, we can validate those drug predictions in our training data set," Brake said.

She added, "We build in the training world, and then we test in the real world. The value of that is that we can tackle real-world problems in the clinical setting, understanding where the breakdown has happened and then hopefully solve, with a drug development partner, how to fix that problem, either in the existing stratification for a therapy or in a second-generation or third-generation drug that's coming along."

In the past three years, the company has focused on building out the engineering stack to support its large dataset and its AI and machine learning tech foundation.

"The build-out now is focused on engaging more in the scientific presence," Brake said. "So, this computational biology and molecular biology that builds out the understanding of where physicians, patients and biopharma can make use of our technology most effectively."

Zephyr AI also is focused on building its commercial footprint to partner with drug developers, all areas where Brake's skills and expertise will be critical to support the company's growth and strategy.

The company plans to use the $111 million cash infusion to enhance its analytical speed and fortify its extensive training and validation data set collection, according to executives. 

"The next step, obviously, is to continue the diversification of our data sources," she noted. "Data comes in different shapes, forms and flavors, and integrating that into a system is very challenging. Doing that comprehensively across a number of different data partners also is very challenging. We have, at Zephyr, spent a lot of time in building that infrastructure, so we have a number of data sources are already available to us, but we're always looking for more, and we're looking to build that proprietary footprint, where we have a knowledge base to give us the opportunity to address as many problems as we possibly can.

Zephyr AI sees big opportunities in the cardiometabolic space to use its datasets and AI algorithms to drive insights into treatments, including newer GLP-1 drugs, and the potential benefits of those medications to a broader series of disease indications, Brake noted.

"We want to validate that data in the real-world setting and we think that there are opportunities for us to engage payers and providers to be able to feed that back into the system to be able to understand how best to utilize those medicines, which patient footprint derives the greatest benefit and to also think about the cost base and the opportunity of what the cost of those medicines are versus the defrayed cost in the medical system," she said.

Drug development is an expensive process, and it can often take researchers years to identify molecules that respond to a given drug—if they’re able to trigger that response at all.

"In biopharma we're typically looking at 12 to 15 years of drug development time and experience because we're historically used to doing this trial and error experimentation in clinics. Patient availability to do those experiments is actually getting harder and harder to access and the cost of patient recruitment is enormous. As a consequence, the time to recruit, and therefore return information that guides and directs your studies, can often be very long and slow," Brake noted. "All the while your IP [intellectual property] is ticking away, and the cost of maintaining this very complex development team to prosecute against those development studies are very enormous."

AI is a disruptive technology that can upend this process, slashing both time and cost. With 20 years spent in drug development, Brake sees this technology disruption as an exciting inflection point for the industry.

"I listen to community physicians all the time, with respect to their challenges of trying to stay abreast of all of this technology, recognizing where molecules are in their development paradigm and the validation of data that they do or don't have," she said. "If we could be more precise with how we enter into drug development, we can shorten our times and we can be more effective in delivery. If we could do that even 10% better than we are, that's an enormous potential that this machine learning technology can empower."

Brake added, "What that means is that we're bringing value, not just for the some of the partners in healthcare that are ultimately prepared to pay for this, but actually we're also bringing benefit to patients because we can bring those medicines to clinic, and ultimately to commercialization, hopefully much faster, in a more effective way. That also then brings this paradigm of value-based healthcare closer to the surface."