How Federated Learning Can Accelerate the Impact of Real-World Data for the Pharmaceutical Industry

Feb 22, 2022

Guest Blog by Yulie Klerman, Vice President of Business Development, Rhino Health

As we enter 2022, precision medicine is – again – high on the priority list for companies across the biopharmaceutical industry. This requires not only access to massive real-world datasets but also advanced artificial intelligence (AI) and machine learning (ML) algorithms that can analyze the data and make it valuable.

While pharmaceutical companies have made strides in utilizing real-world data (RWD) to fuel targeted drug discovery and development in recent years, significant challenges remain. Large-scale studies require the integration of disparate and complex datasets that often have no common language, formatting, or coding. And widely varying, often-changing regulatory requirements across regions and jurisdictions make it difficult to ensure compliance and protect data privacy.

The Federated Opportunity

Unlocking the potential for RWD to deliver on the true promise of precision medicine and accelerate drug discovery and development requires a new approach: end-to-end federated learning and analytics that make it possible to utilize more extensive, more diverse datasets – without ever sharing or aggregating data. Specifically, there are three exciting developments taking shape:

  1. Multimodal AI: Whether focused on mass-market drug discovery or developing precision treatments for a rare disease or cancer, today’s pharma innovators must use multiple types of health data. This requires harmonizing and linking data from different systems and sources (imaging, EMR, pathology lab) while the combined datasets become more vulnerable to privacy violations. Stringent regulations regarding data privacy make gathering and sharing such data within a centralized location practically unattainable. The federated approach unlocks an opportunity at a scale and pace never imagined.
  2. Advanced AI with Federated Learning: Today, too many AI solutions across the healthcare ecosystem fall short of expectations once deployed in the clinical setting. This is because they are created and trained using small datasets that represent only a fraction of the real-world patient population. Medical researchers and AI developers can validate and improve AI models using data from multiple institutions worldwide without ever sharing that data or transferring ownership with federated learning. This opens up unprecedented diversity of data that results in robust AI solutions that work more accurately and more consistently across today’s increasingly diverse patient populations.
  3. Federated RWD Technology: For years, industry leaders have been talking about real-world data (RWD) potential to revolutionize every clinical research stage – from trial design to patient identification to participant selection to outcomes measurement. Now, with federated analytics and AI models running at the edge (via a federated approach that eliminates the need to move data from a central repository), this vision is coming to fruition. Clinical trial teams are able to utilize deeper, more valuable clinical data – well beyond the data many have had to rely on previously.

Transforming Clinical Trials With Federated RWD Platform

With these possibilities powered by a federated approach to accessing and utilizing RWD, the pharmaceutical industry is well-equipped to transform clinical trials fundamentally. First, through faster and more efficient recruitment – including tapping into hard-to-find patient pools. Leveraging distributed real-world datasets and mining more across multimodal data in real-time will help researchers meet enrollment targets quickly and effectively. Then, applying AI models that identify the most relevant cohorts (particularly those with rare diseases or cancers with specific genetic mutations) they can inform their healthcare providers of the trial opportunity.

Second, embracing the federated approach will help expand clinical trials through synthetic control arms. None of the aggregated data repositories can currently mine a sufficient number of patients answering the protocol criteria, but federated access to RWD can break this constraint by opening up access to a substantial amount of data worldwide. This has a significant potential benefit to patients, as creating a synthetic control arm means that all participants in a trial will receive the drug – not the placebo.

Finally, the federated approach can improve outcome measurements. The ability to better analyze significantly larger real-world datasets helps researchers identify and understand novel indications or side effects that may not be revealed in any one clinical trial due to relatively small numbers of participants. Also, the ability to integrate data from multiple sources and systems provides a broader window into patients’ health, entitling researchers and HCPs to predict disease onset better and offer more informed treatment recommendations. The bottom line is that precision medicine at scale is not feasible without robust AI, and robust AI can only be trained with massive RWD data. 2022 promises to be a breakthrough year in utilizing RWD in the biopharmaceutical industry – fueled by a federated approach that facilitates data standardization and harmonization, unleashes access to distributed datasets, and orchestrates AI workflows across medical sites around the world, all while safeguarding privacy.


Yulie Klerman
Vice President of Business Development, Rhino Health

Yulie is a VP of Business Development at Rhino Health, an end-to-end Federated Learning and Privacy Preservation platform enabling privacy-centric clinical data collaboration. 

Previously she led the New Ventures & Business Incubation function at Liveramp Health. She evaluated disruptive technologies, managed strategic venture partnerships & investments, built out innovation capabilities, and commercialized new services. She launched a first-in-class Health Data Connectivity platform for the safe and effective incorporation of health, consumer, and digital data. 

Yulie is obsessed with data interoperability and connectivity. For the last 13 years, Yulie has been living and breathing healthcare innovations, bringing solutions to healthcare stakeholders around the globe. She demonstrates a strong track record in building strategic partnerships at Alma Lasers, Mazor Robotics (acquired by Medtronic), and Philips Healthcare. 

Yulie graduated from Weizmann Institute of Science with an M.Sc in Physical Chemistry. She currently serves on the Advisory Board of iFocus Health, an eye-tracking platform integrated with a proprietary AI algorithm enabling a data-driven ADHD treatment.

See all MassBio News