Logistic regression, Cox proportional hazards, Kaplan-Meier, log-rank test, chi-square test, two-sample t-test and summary tables (Table 1). The results match a pooled analysis of all sites' data.

Row-level data never leaves the data holder. Each site runs the cohort, the dataset and the analysis on its own records and returns aggregate statistics only.

Before anything leaves, each site applies its own minimum-count floors, 5 per cell and 10 per site by default. A requesting site can make them stricter but cannot loosen them.

Access between sites runs through contract agreements between data connectors. Connectors authenticate each other, and every result file is cryptographically signed.

Federated studies run inside a Secure Processing Environment, a remote workspace you open in the browser. Teams sign in with single sign-on and share projects.