Abstract

Objective

The objective of this study was to establish a nomogram model to predict SI in patients with cancer and further evaluate its performance.

Method

This study was performed among 390 patients in oncology departments of Affiliated Hospital of Nantong University from April 2020 to January 2021. Of these, eligible patients who were diagnosed with cancer were split into training and validation cohorts according the ratio of 2:1 randomly. In the training cohort, multivariate regression was performed to determine the independent variables related to SI. A nomogram was built incorporating these variables. The model performance was evaluated by an independent validation cohort.

Results

The prevalence of SI in patients with cancer was 22.31% and 19.23% in training and validation cohorts, respectively. The nomogram model suggested independent variables for SI, including depression, emotional function, time after diagnosis, family function and educational status. The area under the curve (AUC) was 0.93 (95%CI, 0.90–0.97) and 0.82 (95%CI, 0.74–0.90) in training and validation cohorts respectively, which indicated good discrimination of the nomogram in predicting SI in cancer patients. The p-value of the goodness of fit (GOF) test was 0.197 and 0.974 in training and validation cohorts respectively, suggesting our nomogram model has acceptable calibration power, and the calibration curves further indicated good calibration power.

Conclusion

In conclusion, the nomogram model for predicting individualized probability of SI could help clinical caregivers estimate the risk of SI in patients with cancer and provide appropriate management.

中文翻译：

---

开发和验证预测癌症患者自杀风险的模型

摘要

客观的

本研究的目的是建立一个列线图模型来预测癌症患者的 SI 并进一步评估其性能。

方法

本研究于 2020 年 4 月至 2021 年 1 月对南通大学附属医院肿瘤科的 390 名患者进行了研究。其中，符合条件的诊断为癌症的患者按照 2:1 的比例随机分为训练队列和验证队列。在训练队列中，进行多元回归以确定与 SI 相关的自变量。构建了包含这些变量的列线图。模型性能由一个独立的验证队列进行评估。

结果

在训练和验证队列中，癌症患者的 SI 患病率分别为 22.31% 和 19.23%。列线图模型建议 SI 的自变量，包括抑郁、情绪功能、诊断后时间、家庭功能和教育状况。曲线下面积 (AUC) 在训练和验证队列中分别为 0.93（95%CI，0.90-0.97）和 0.82（95%CI，0.74-0.90），这表明列线图在预测癌症患者 SI 方面具有良好的辨别力. 拟合优度 (GOF) 检验的 p 值在训练和验证队列中分别为 0.197 和 0.974，表明我们的列线图模型具有可接受的校准能力，校准曲线进一步表明良好的校准能力。

结论

总之，用于预测 SI 个体化概率的列线图模型可以帮助临床护理人员估计癌症患者 SI 的风险并提供适当的管理。