Development and Validation of a Nomogram for Predicting the Unresolved Risk of Parents of Adolescents With Psychiatric Diagnoses

Abstract

Evaluating the resolution of parents of ill children can help in taking measures to alleviate their distress in a timely manner and promote children's rehabilitation. This study aims to develop and validate a nomogram for predicting the unresolved risk of parents of adolescents with psychiatric diagnoses. The data for 130 parents (modeling dataset = 90; validation dataset = 40) were collected. A nomogram was first developed to predict the unresolved risk for parents based on the logistic regression analysis in the modeling dataset. The internal and external validation then were conducted through quantifying the performance of the nomogram with respect to discrimination and calibration, respectively, in the modeling and validation datasets. Finally, the clinical use was evaluated through decision curve analyses (DCA) in the overall dataset. In the results, the nomogram consisted of six risk factors and provided a good discrimination with areas under the curve of 0.920 (95% CI, 0.862-0.978) in internal validation and 0.886 (95% CI, 0.786-0.986) in external validation. The calibration with good consistency between the observed probability and predicted probability was also found in both internal and external validation. DCA showed that the nomogram had a good clinical utility. In conclusion, the proposed nomogram exhibited a favorable performance with regard to its predictive accuracy, discrimination capability, and clinical utility, and, thus, can be used as a convenient and reliable tool for predicting the unresolved risk of parents of children with psychiatric diagnoses.

Keywords: nomogram; parents; prediction model; psychiatric diagnoses; resolution; unresolved risk.

Figure 1

Nomogram for predicting the unresolved risk of parents of children with psychiatric diagnoses. Each risk factor of the parents has a value on its variable axis, and vertical lines are drawn upward to the point axis to obtain the corresponding scores for the value of each risk factor. A total score can be easily calculated by adding each single score. The total score is plotted on the bottom total point axis, and a vertical line is drawn downward to the unresolved risk axis to estimate the unresolved risk.

Figure 2

The calibration plot of the nomogram in the modeling dataset. The x-axis shows the nomogram-predicted probability of unresolved risk, and the y-axis shows the actual observed probability of unresolved risk. The ideal line (45°) in the calibration plots represents perfect consistency between the nomogram-predicted probabilities and the observed probabilities. The apparent line represents the model calibrated with the modeling dataset, and the bias-corrected line represents the model calibrated using the bootstrap method.

Figure 3

The receiver operating characteristic (ROC) of the nomogram in the modeling dataset.

Figure 4

The calibration plot of the nomogram in the validation dataset. The x-axis shows the nomogram-predicted probability of unresolved risk, and the y-axis shows the actual observed probability of unresolved risk. The ideal line (45°) in the calibration plots represents perfect consistency between the nomogram-predicted probabilities and the observed probabilities. The apparent curve represents the model calibrated with the validation dataset, and the bias-corrected curve represents the model calibrated using the bootstrap method.

Figure 5

The receiver operating characteristic (ROC) curves of the nomogram in the validation dataset.

Figure 6

The decision curve analysis of the nomogram in the entire dataset. The y-axis measures the net benefit, and the x-axis measures the probability threshold. The red line represents the nomogram. The black line (treat-none scheme line) represents the assumption that all parents are unresolved statuses. The gray line (treat-all-parents scheme line) represents the assumption that no parent is unresolved status. The decision curve analysis (DCA) indicated that the nomogram provided more benefits than the treat-all-parents scheme and treat-none scheme when the threshold probability of unresolved status was >13% and <54%.