Join top executives in San Francisco on July 11-12, to hear how leaders are integrating and optimizing AI investments for success. Learn More
The potential for artificial intelligence (AI) and machine learning (ML) to improve human health cannot be understated, but it does face challenges.
Among the big challenges is dealing with siloed data sources, so researchers are not able to easily analyze data from multiple locations and initiatives, while still preserving privacy. It’s a challenge that can potentially be solved with an approach known as federated learning.
Today in a research report first published in Nature Medicine, AI biotech vendor Owkin has revealed just how powerful the federated model can be for healthcare. Owkin working alongside researchers at four hospitals in France was able to build a model with its open source technology that it claims will have a significant impact on the ability to help effectively treat breast cancer. The Owkin AI models were able to identify accurately novel biomarkers that could lead to improved personalized medical care.
“Owkin is an AI biotech company and we really have this ambitious goal, which is to cure cancer,” Jean du Terrail, senior machine learning scientist at Owkin, told VentureBeat. “We are trying to leverage the power of AI and machine learning, in addition to our network of partners, to move towards this goal.”
Join us in San Francisco on July 11-12, where top executives will share how they have integrated and optimized AI investments for success and avoided common pitfalls.
Owkin is one of the hottest biotech startups in the market today. The company raised $80 million in funding back in June 2022, from pharmaceutical giant Bristol Myers Squibb, bringing total funding to the unicorn startup, over $300 million since the company was founded in 2016.
Why federated learning is critical for the advancement of AI healthcare
In healthcare and clinical studies, there is often a significant amount of personally identifiable information that needs to be protected and kept private. Researchers as well as hospitals will also often be required to keep some data within their own organizations, which can lead to information silos and collaboration friction.
Terrail explained that federated learning provides an approach by which ML training can occur across the different information silos on patient data located in hospitals and research centers. He emphasized that the approach that Owkin has developed does not require that data ever actually leaves the source facility and patient privacy is protected.
The federated learning approach is an alternative to using synthetic data, which is also commonly used in healthcare to help protect privacy. Terrail explained that federated learning enables researchers to access real world data that is secured behind firewalls and is often difficult to access. In contrast, synthetic data is simulated data that potentially may not be entirely representative of what can be found in the real world. The risk with synthetic data in Terrail’s view is that AI algorithms built with it could potentially not be accurate.
To protect patient privacy, the Owkin approach involves having data going through a process known as pseudonymization. Terrail explained that the pseudonymization process basically removes any personally identifiable information.
The open source software that enables federated learning
Owkin developed a technology stack for federated learning called Substra, that is now open source. The Substra project is currently hosted by the Linux Foundation’s AI and Data Initiative.
Terrail said that the Substra platform enables data engineers in hospitals to connect sources remotely for the ML training. He referred to Substra as a ‘PyTorch on steroids’ application that enables researchers to add capabilities on top of existing machine learning frameworks, such as PyTorch. The additional capabilities enable the federated learning model approach, where data is located securely and privately in disparate locations.
The Substra technology also makes use of the open source Hyperledger immutable ledger blockchain technology. The Hyperledger technology enables Substra and Owkin to be able to accurately track all the data that is used. Terrail said that Hyperledger is what enables traceability into every operation that is done with Substra, which is critical to ensuring the success of clinical efforts. With traceability, researchers can verify all the steps and data that was used. Additionally it helps with enabling interpretable AI as the data doesn’t all just reside in a black box that no one can audit.
Improving breast cancer treatment with federated learning
The Owkin teams worked with researchers across four hospitals, and were able to train the federated learning model on clinical information and pathology data from 650 patients.
“We trained the model to predict the response of the patient to neoadjuvant chemotherapy, which is the gold standard,” Terrail said. “It’s basically what you give to triple negative breast cancer patients that are in the early stage, but you don’t know if it is going to work or not.”
The research was designed to build an AI that could determine how a patient will respond and whether or not the treatment is likely to work. The model could also help to direct a patient to other treatments.
The cancer treatment breakthrough according to Thomas Clozel, co-founder and CEO of Owkin is predicated on the success of the federated learning model that is able to gather more data to train the AI than what had been done previously.
“We want to build federated learning to break competitive and research silos,” Clozel told VentureBeat. “It’s about human connection and being able to really create this federated network of the best practitioners in the field and researchers being able to work together.”
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.