Analyzing factors influencing hospitalization costs for five common cancers in China using neural network models

Malignant tumors are a major global health crisis, causing 25% of deaths in China, with lung, liver, thyroid, breast, and colon cancers being the most common. Understanding the factors influencing hospitalization costs for these cancers is crucial for public health and economics. This study aimed to identify key cost factors and develop a neural network model for predicting hospitalization costs, thereby providing tools to ease the financial burden on patients and healthcare systems. Data on hospitalization costs for 30,893 cancer patients from secondary or higher-level hospitals in Zhuhai, Guangdong Province, between 2017 and 2022, were analyzed. Neural network classification and feature importance analysis were used to determine the main factors influencing costs and to develop predictive models. Model performance was evaluated using the area under the receiver operating characteristic curve (AUROC), with a 95% confidence interval (CI) calculated for the AUROC value. The key factors influencing hospitalization costs for lung cancer are metastasis and malignant solid tumor (MST), with correlation coefficients of 0.126 and 0.086, respectively, both showing statistical significance (p p This study confirmed the strong clinical applicability of the neural network predictive model in analyzing hospitalization costs for lung and colon cancer and revealed the factors that influence hospitalization costs for these cancers.

恶性肿瘤是全球性重大公共卫生危机，在中国其导致的死亡占总死亡人数的25%，其中肺癌、肝癌、甲状腺癌、乳腺癌与结直肠癌为最常见的癌种。明确上述癌症患者住院费用的影响因素，对公共卫生与经济学研究均具有重要价值。本研究旨在识别关键成本影响因素，并开发用于预测住院费用的神经网络（neural network）模型，以期为减轻患者与医疗系统的财务负担提供实用工具。本研究分析了广东省珠海市2017年至2022年间，二级及以上医院收治的30893名癌症患者的住院费用数据。研究采用神经网络分类与特征重要性分析方法，确定影响住院费用的主要因素并构建预测模型。以受试者工作特征曲线下面积（AUROC）评估模型性能，并计算AUROC值的95%置信区间（CI）。影响肺癌患者住院费用的关键因素为转移与恶性实体瘤（MST），相关系数分别为0.126与0.086，二者均具有统计学显著性（p p）。本研究证实了神经网络预测模型在分析肺癌与结直肠癌住院费用方面具备良好的临床适用性，并揭示了上述癌症住院费用的相关影响因素。