基于决策树的成人斜视患者生存质量预测模型研究

Prediction model of quality of life of adult strabismus patients based on decision tree

目的探讨决策树在成人斜视患者生存质量预测模型中的应用价值。方法收集618例成人斜视患者的人口学资料、眼科专科情况以及社会心理指标,建立成人斜视患者生存质量多元线性回归模型和决策树模型,并将样本按3∶1的比例分为训练集及测试集。结果多元线性回归模型中抑郁、斜视类型、消极应对、斜视度数、焦虑和弱视是重要影响因素(均P<0.01),其预测正确率为73.62%;决策树模型中抑郁、年龄、斜视类型和斜视度数是重要影响因素,其正确率为92.24%。结论决策树模型较多元线性回归方程更能较好预测成人斜视患者生存质量,更有利于及时了解患者生存质量变化,从而制定更具针对性的干预措施。

Objective To explore the value of decision tree in predicting the quality of life of adult strabismus patients. Methods Demographic data,ophthalmologic clinical data and social psychological indicators were collected from 618 adult patients with strabismus. The predicting model of quality of life among adult strabismus was built by multiple linear regression model and decision tree model. Results By stepwise regression analysis, such factors as depression, types of strabismus, negative coping, strabismus degree, anxiety and amblyopia were screened out（P〈0.01 for all）, with correct predicting rate being 73.62%. In the decision tree model, depression, age, strabismus type and strabismus degree were important factors influencing quality of life of adult strabis mus patients, with correct predicting rate being 92. 24%. Conclusion Decision tree model can better predict the quality of life of adult strabismus patients; it helps foster a timely understanding of quality of life of patients, and establish intervention measures for patients.