The World Economic Forum's MINDS programme has spotlighted FLock.io for its privacy-preserving AI collaborations with Moorfields Eye Hospital and University College London Hospitals. By utilizing federated learning, these institutions train clinical models for eye disease detection and diabetes management without sharing sensitive patient data. FLock.io estimates that a 1% reduction in the NHS's £10B+ annual diabetes budget could save over £100M annually.
FLock.io World Economic Forum recognition
- ▪The World Economic Forum's MINDS cohort includes organizations such as Lenovo, Occidental, TCL Industries, Hisense Hitachi, and KUKA
- ▪The World Economic Forum's MINDS programme spotlighted FLock.io for its privacy-preserving AI work with two National Health Service trusts
NHS federated learning use cases
- ▪FLock.io is working with researchers from University College London and University College London Hospitals to develop locally trained glucose monitoring alerts using data from over 400 patients
- ▪Moorfields Eye Hospital and University College London Hospitals are using FLock.io's federated learning platform for eye disease detection and diabetes management
Privacy-preserving AI technology
- ▪The government of Sarawak, Malaysia is completing a sovereign AI pilot with FLock.io to establish standards for cross-border healthcare AI collaboration
- ▪Federated learning allows collaborative AI model training by keeping raw data locally on-premises or on edge devices and sharing only encrypted model updates
Healthcare AI cost savings
- ▪FLock.io estimates that AI-driven prevention in the National Health Service could save over £100 million annually based on a 1% reduction in diabetes management costs
- ▪The National Health Service currently spends over £10 billion annually on diabetes management
FLock.io platform performance metrics
- ▪FLock.io's architecture combining federated learning and blockchain-based verification achieves a 37% improvement in model accuracy and a 44% reduction in total cost of ownership
- ▪FLock.io's platform reduces deployment time by 63% and uses 80% less training energy per model update
Story comments
Loading comments…