Algorithm: ODAL
1. Initial value: obtain $\overline{\beta }=\text{arg\hspace{0.17em}}\underset{\text{β}}{\text{max}}{L}_1\left(\beta \right).$ using data in the local site (i.e., site 1), where L1(β) is the log likelihood of logistic regression defined in equation (2)
2. Initial communication: transfer $\overline{\beta }$ to the other sites (i.e., sites 2, 3, …, K)
3. For j = 2 to K
4.  do compute $\nabla L_j\left(\overline{\beta }\right)$, where Lj(β) is defined similarly as in equation (2)
5.   transfer $\nabla L_j\left(\overline{\beta }\right)$ to the local site
6.  end
7. Compute the surrogate likelihood $\tilde{L}\left(\beta \right)$ defined in equation (3)
8. obtain $\tilde{\beta }=\text{arg\hspace{0.17em}}\underset{\text{β}}{\text{max}}\tilde{L}\left(\beta \right)$
9. return $\tilde{\beta }$