This article introduces extensions to decision curve analysis (DCA), a novel method for evaluating diagnostic tests, prediction models, and molecular markers. DCA combines the simplicity of accuracy measures like sensitivity and specificity with the clinical applicability of decision analytic approaches. Key extensions include: 1. **Correction for Overfit**: Repeated 10-fold cross-validation is recommended to correct decision curves for overfitting, providing the best method among various techniques. 2. **Confidence Intervals**: Bootstrap methods are proposed to calculate confidence intervals for net benefits, enhancing the interpretability of decision curves. 3. **Application to Censored Data**: The method is extended to handle survival time data subject to censoring, ensuring accurate evaluation of models in clinical settings. 4. **Direct Calculation from Predicted Probabilities**: Decision curves can be calculated directly from predicted probabilities, smoothing the curve and improving reliability. The authors also provide software to implement these extensions, making DCA more accessible for practical use. The extensions address limitations in traditional DCA, such as overfitting, lack of confidence intervals, and the need for complete data sets. The article includes simulation studies and real-world examples to demonstrate the effectiveness of these extensions.This article introduces extensions to decision curve analysis (DCA), a novel method for evaluating diagnostic tests, prediction models, and molecular markers. DCA combines the simplicity of accuracy measures like sensitivity and specificity with the clinical applicability of decision analytic approaches. Key extensions include: 1. **Correction for Overfit**: Repeated 10-fold cross-validation is recommended to correct decision curves for overfitting, providing the best method among various techniques. 2. **Confidence Intervals**: Bootstrap methods are proposed to calculate confidence intervals for net benefits, enhancing the interpretability of decision curves. 3. **Application to Censored Data**: The method is extended to handle survival time data subject to censoring, ensuring accurate evaluation of models in clinical settings. 4. **Direct Calculation from Predicted Probabilities**: Decision curves can be calculated directly from predicted probabilities, smoothing the curve and improving reliability. The authors also provide software to implement these extensions, making DCA more accessible for practical use. The extensions address limitations in traditional DCA, such as overfitting, lack of confidence intervals, and the need for complete data sets. The article includes simulation studies and real-world examples to demonstrate the effectiveness of these extensions.