Open Source Platform for Health AI Launched

The new open-source platform, named FLIP (Federated Learning Interoperability Platform), was launched by the AI Centre for Value-Based Healthcare and deepc. This initiative aims to allow AI models to be trained across multiple institutions while the data remains in secure, local environments, a process known as federated learning. Key partners in this NHS-led resource include King's College London, Guy's and St Thomas' NHS Foundation Trust, OneLondon, and Flower Labs. This federated approach directly confronts the problem of data silos in healthcare, which have been a major obstacle to developing accurate and generalizable AI models. Training AI on fragmented datasets from single institutions can lead to biased and underperforming models, costing organizations millions and posing risks to patient safety. The global healthcare economy loses an estimated $3.1 trillion annually due to such inefficiencies. The platform's architecture supports standards like FHIR, SNOMED, and DICOM, which are crucial for interoperability between different electronic health record (EHR) systems. This focus on standardization is critical, as inconsistent data formats and a lack of interoperability are persistent challenges in healthcare IT, hindering the creation of comprehensive patient views. Composable architecture, a core principle of the new platform, allows healthcare organizations to adopt modular solutions rather than monolithic EHR systems. This approach enables the integration of best-of-breed applications and facilitates more rapid adaptation to new technologies and clinical needs. It reflects a broader industry shift towards open, data-centric platforms that separate the data layer from the application layer. The use of open-source software in healthcare, while historically approached with caution, is gaining momentum as a way to accelerate innovation and reduce dependency on proprietary systems. Projects like MONAI for medical imaging and platforms from companies like Tuva Health have demonstrated the power of community-driven development in creating specialized, effective tools. This initiative is part of a larger trend toward applying advanced AI and large language models (LLMs) to solve complex healthcare challenges. The global AI healthcare market is projected to reach $187 billion by 2030, with 63% of healthcare professionals already actively using AI in 2025. Companies like Google, with its MedGemma model, and startups funded by Y Combinator are actively developing AI-native tools for everything from clinical documentation to insurance denial automation. By providing an open, secure, and interoperable framework, FLIP aims to enable researchers and clinicians to harness data across institutions safely. An early project is already underway with St John's Institute of Dermatology, developing digital biomarkers for inflammatory skin disease using multimodal data, showcasing the platform's potential for real-world model development.