Aim 3: federated meta-learning of AE Risks

Improve the generalizability of RWE on AE risks through federated meta-learning of multi-institutional EHR (multi-EHR) data.

Synthesizing information from multiple sources across different health systems has the potential to greatly enhance the generalizability of RWE, especially for AE detection. We will :

• Quantify AE risks associated with DMTs for a given target population via federated robust meta-learning with multi-EHR data;
• And develop federated robust estimation of conditional treatment effect estimation.
• We will perform real-world studies assessing the risk of serious infection from DMTs using multi-EHR data for RA and MS.