Case Study: Tel Aviv (Sourasky) Medical Center improves NICU length-of-stay prediction with Rhino Federated Computing

Federating a Quantile Regression Model for Neonatal Care at Tel Aviv (Sourasky) Medical Center Using Rhino Health’s Federated Computing Platform

The Tel Aviv (Sourasky) Medical Center sought to improve neonatal care by predicting the length of stay (LOS) for premature newborns in its NICU. This required a predictive model that could leverage diverse data while strictly adhering to patient privacy regulations like HIPAA and GDPR. The challenge was to develop this model without centralizing sensitive patient information from different sites, a common barrier in healthcare collaboration. Rhino Federated Computing provided its Federated Computing Platform (FCP) as the solution to this challenge.

Rhino Federated Computing implemented a Federated Learning Quantile Regression model on its FCP, which trained the algorithm across decentralized data sources without moving sensitive patient information. The federated model maintained performance parity with a traditional centralized model, achieving an R-squared value of 0.787, validating that it could deliver accurate predictions without compromising data privacy. This successful proof of concept demonstrated the platform's ability to facilitate secure, multi-institutional collaborations, paving the way for more effective and inclusive healthcare AI.