Fig. 5
From: Survival analysis using machine learning in transplantation: a practical introduction
Decision curve analysis. The decision curve plot illustrates the net benefit of using the predictive model across various threshold probabilities. The model curve indicates the net benefit of the predictive model at different threshold probabilities. The model curve consistently lies above the "Treat All" and "Treat None" curves, suggesting that the model provides a higher net benefit than either treating all subjects or treating none. The threshold probability at which the model curve reaches its peak net benefit can be considered the optimal threshold for decision-making. This point represents the maximum clinical utility of the model. By comparing the model curve with the "Treat All" and "Treat None" curves, we can assess the model's improvement over these baseline strategies. The model demonstrates a significant net benefit over both baselines, indicating its potential value in clinical practice