Algorithm 2. Action selection

Input: $S_t^g$, $P_{gt}$

Output: A patient with maximum $Q$-value

1: rn $\leftarrow$ random ( ):

2:  if rn $\le ϵ$ then

3:   action $\leftarrow$ random $\left(P_{gt}\right)$

4:  else

5:   action $\leftarrow$ maxQvalue $\left(S_t^g,P_{gt}\right)$

6:  end if