Fig. 4

Construction and validation of the depression prediction model for middle-aged and elderly physically disabled people in China (A) nomogram: the sum of the scores on each predictor, which predicts the probability that depression will occur (B) ROC plot of the model: the horizontal coordinate is the false positive rate, representing the proportion of false positive samples. The vertical coordinate is the sensitivity, representing the proportion of true positive samples. The two form the ROC curve, which is used to assess the model’s ability to correctly classify positive and negative samples at different classification thresholds (C) Calibration curves: the horizontal coordinate is the probability of an event occurring as predicted by the model. The vertical coordinate is the proportion of events that actually occur within the predicted probability range. The calibration curve is used to assess the agreement between the predicted probability of an event and the actual probability of its occurrence (D) DCA chart: the horizontal coordinate is the high risk threshold, referring to the thresholds selected for the different predictor variables. The vertical coordinate is the standardised net benefit, which refers to the standardised net benefit calculated for the different predictor variables constituting the model and the line strategy. The DCA plot of the two makes it possible to assess the extent to which each model outperforms or underperforms the baseline strategy under different decision scenarios (E) Net Benefit Curve: The high-risk thresholds and cost-benefit ratios in the horizontal coordinate compare the costs of using the model with its benefits. The vertical coordinate represents the number of samples judged to be depressed at the selected high-risk threshold for a sample size of 1,000. The three form a net benefit curve that can be used to assess the benefits of each model across different predictor variables, helping policymakers to make optimal decisions