Apheris Uses Federated Computing to Tackle Life Science Data Challenges

In the field of life sciences, particularly in the pharmaceutical industry, one of the core challenges in building AI solutions is the utilization of data. Although AI development heavily relies on data, much of the health data remains underutilized, primarily due to patient privacy protection, regulatory restrictions, and intellectual property concerns.

To address this challenge, German entrepreneur Robin Röhm founded the startup Apheris, which employs federated computing technology. This decentralized approach allows data to be used securely for AI model training without moving it. By performing computations locally where the data resides and only aggregating the outputs, such as model parameters, Apheris ensures data security. The company's clients include major enterprises like Roche and several hospitals.

According to Marcin Hejka, co-founder and managing partner of OTB Ventures, Apheris has the potential to become a key component in the emerging federated data network. As the ecosystem of third-party software tools matures, Apheris also supports seamless integration with complementary privacy-enhancing technologies, such as homomorphic encryption, differential privacy, and synthetic data.

Initially established in 2019, Apheris aimed to create a federated learning framework that could compete with open-source methods. However, after securing seed funding in 2022, the company shifted its focus in 2023 to cater to data owners, particularly in the pharmaceutical and life sciences sectors. This pivot led to a better product-market fit, resulting in a fourfold increase in revenue since the launch of their new product in the fourth quarter of 2023.

The new funding will be used to hire senior talent with backgrounds in life sciences, including those in commercial roles. Apheris Compute Gateway, a software agent that connects local data with AI models, has been adopted by the AI-Driven Structural Biology Alliance, which includes members such as AbbVie, Boehringer Ingelheim, Johnson & Johnson, and Sanofi, all working together on AI-driven drug discovery.

Furthermore, Apheris plans to focus more on areas like protein complex prediction, recognizing the additional value it can provide when publicly available data is limited. The company is committed to addressing the concerns of data owners regarding sharing data with AI, as it believes that the true impact of AI cannot be realized without resolving these issues.