基于神经网络模型的青年癌症患者心理痛苦预警与标准化沟通系统研究

Today, as psychological problems are getting increasingly serious, psychological communication will become a powerful means to treat cancer. Based on the Distress Thermometer, Hospital Anxiety and Depression Scale, and Medical Coping Modes Questionnaire, the investigation found that the psychological distress occurrence rate of young cancer patients is 89.1%, and the anxiety and depression rates are 90% and 75% respectively, which are higher than data in domestic reports. Meanwhile, young cancer patients’ abilities to coping with the situation such as facing the challenge and the avoidance dimension score was significantly lower than that of reported. This dissertation, using the existing sample database and Xiangya cancer big data resources, on the one hand, explore the features of young cancer patients on psychological distress, risk factors and potential consequences through qualitative research, analyze the correlation, and then establish the BP neural network prediction model based on the integrated preliminary data, and comparing with the logistic regression model, decision tree model (C5.0). On the other hand, combinated with the predicative analysis of neural network, it can find the accurate early warning object, implementing pre-warning intervention and communication intervention respectively to patients who have clinical psychological distress or no-apparent-distress-yet patients, then according to the data collected before and after intervention, using BP neural network, eventually construct a psychological distress communication intervention model for young cancer patients, which can help it knows whether cancer patients need to receive communication intervention for decision support, validation and further improve the early warning and intervention system communication.

心理问题日益严重。心理沟通将成为癌症治疗的有力手段。前期通过运用心理痛苦温度计、医院焦虑抑郁量表、医学应对问卷等对青年癌症患者进行调查，研究发现其心理痛苦发生率为89.1%，焦虑、抑郁发生率分别为90%、75%，均高于国内报道；青年癌症患者医学应对情况差，其面对、回避等维度得分明显低于国内报道。本课题将利用现有样本库和湘雅肿瘤相关大数据资源，一方面通过定性研究，探寻青年癌症患者心理痛苦的表现特征、危险因素及潜在后果，并分析其相关性，综合数据建立BP神经网络预测模型，并与Logistic回归模型、决策树C5.0模型等进行比较。另一方面，结合预测模型分析，准确发现预警对象，分别对未出现和已出现心理痛苦的青年癌症患者实施预警干预和沟通干预，并根据沟通干预前的基线数据和干预后的随访数据，利用BP神经网络，构建青年癌症患者心理痛苦沟通干预的决策模型，为青年癌症患者是否需要接受沟通干预进行决策支持。

本课题利用现有样本库和湘雅肿瘤相关大数据资源，一方面通过定性和定量研究，探寻青年癌症患者心理痛苦的表现特征、危险因素及潜在后果，并分析其相关性，综合数据建立BP神经网络预测模型，并与Logistic回归模型、决策树C5.0模型等进行比较。另一方面，结合预测模型分析，准确发现预警对象，对青年癌症患者实施预警和沟通干预，并根据沟通干预前的基线数据和干预后的随访数据，为青年癌症的沟通干预提供决策支持。