Apheris: Federated Computing for Life Science AI Data
The Challenge of Data Access in AI for Life Sciences
Artificial intelligence relies heavily on data, yet a significant portion of valuable health data remains untapped. This is primarily due to concerns surrounding patient privacy, stringent regulations, and the protection of intellectual property.
Robin Röhm, a German entrepreneur, identifies this as the “core underlying problem” hindering the development of AI solutions within the life sciences and pharmaceutical industries. Furthermore, collaborative efforts involving sensitive data often present considerable obstacles.
Apheris and Federated Computing
Röhm’s startup, Apheris, proposes a solution through federated computing. This innovative approach enables secure access to data for AI model training without the need for data transfer, adopting a decentralized methodology.
The company currently serves clients including Roche and numerous hospitals, demonstrating the practical application of its technology.
How Federated Computing Works
Marcin Hejka, a co-founder and managing partner at OTB Ventures, explains that federated computing centers on the principle that “computations are executed locally where data resides.” Only the resulting outputs, such as model parameters, are then aggregated centrally.
Hejka believes Apheris is poised to become a vital element within the emerging federated data networks. He notes the growing maturity of supporting tools, including open-source federation engines, data quality solutions, and security products.
Apheris also facilitates integration with complementary privacy-enhancing technologies like homomorphic encryption, differential privacy, and synthetic data generation.
From Federated Learning Framework to Data Owner Focus
Apheris initially began in 2019 with the aim of developing a federated learning framework to compete with open-source alternatives. This was based on the founders’ prior experience at Janus Genomics.
Following a substantial seed funding round in 2022, the company underwent a significant strategic shift in 2023. The focus was redirected towards empowering data owners and concentrating on the pharmaceutical and life sciences sectors.
This pivot proved successful, with the startup achieving product-market fit with its latest offering launched in late 2023. Revenue subsequently increased fourfold.
Recent Funding and Future Plans
Backed by investors including Octopus Ventures and Heal Capital, Apheris has secured $8.25 million in Series A funding, bringing its total funding to $20.8 million. This capital will be used to recruit experienced professionals, particularly those with backgrounds in life sciences and commercial roles.
The Apheris Compute Gateway, the software agent facilitating communication between local data and AI models, is currently utilized by the AI Structural Biology (AISB) Consortium. This consortium includes prominent members such as AbbVie, Boehringer Ingelheim, Johnson & Johnson, and Sanofi, who are collaborating on AI-driven drug discovery.
Focus on Protein Complex Prediction
With this new funding, Apheris will further concentrate on protein complex prediction. The company recognizes its ability to deliver value in scenarios where publicly available data is limited, but substantial, diverse data remains inaccessible due to security concerns.
“Without addressing the data owners’ concerns in providing data to AI, we don’t think that the impact of AI can really be unlocked,” Röhm stated, emphasizing the core mission of Apheris.
Stay informed with TechCrunch’s AI newsletter! Subscribe here to receive it weekly in your inbox.
