Agentic data science with the Data Agent Kit
cloud.google.com·7h ago
AI Research
arXiv CS.AI·Aug 11·5 min read
Communication-efficient distributed hazard difference estimation for heterogeneous multi-site survival data
TL;DR
DiSAH is a novel federated algorithm for estimating hazard differences in survival data without sharing patient-level information. It matches centralized analysis in accuracy while enabling collaboration across multiple sites.
✦ Why It Matters
Implement DiSAH to enhance survival analysis capabilities while maintaining patient data privacy in your healthcare organization.
Key Takeaways
- Utilize DiSAH for privacy-preserving survival analysis across multiple sites.
- Aggregate summary statistics instead of sharing patient-level data.
- Leverage DiSAH to identify risk factors not detectable by individual sites.
- Apply findings for better resource allocation and health-economic evaluations.
How It Works
DiSAH employs a non-iterative, closed-form approach to estimate hazard differences, which represent the absolute change in event rates due to risk factors. By focusing on summary statistics, it eliminates the need for a central server, allowing each participating site to perform necessary calculations independently while ensuring patient data remains secure.