Compute the classification accuracy as the metrics for model selection.

• Select the best model with the highest accuracy from the above grid search, and calculate the IPT weight by transforming the estimated propensity score.

**Step 2.** Assess the treatment effects using the IPT weighted survival analysis.

• Construct a multilayer perceptron with two hidden layers, ReLU activations, and the proposed loss function l(θ).

• For a fixed learning rate, the number of nodes in hidden layers from a grid of hyper-parameter values: # Compile the model using Adam optimizer.
