基于机器学习技术的在线疾病诊疗方案倾向性识别研究

Machine learning technique-based tendency cognition of online disease diagnosis and treatment plan

目的:采用机器学习技术分析和预测在线疾病诊疗方案的倾向性。方法:爬取39疾病百科网中与心血管疾病相关的439条症状数据,通过TF-IDF算法提取症状关键词向量,采用支持向量机、决策树、神经网络建立分类模型,预测不同病症倾向西医或中西医结合的治疗方案。结果:对在线疾病信息的文本分析可挖掘疾病的特征,采用机器学习技术可有效预测对应治疗方案的倾向性,预测精度均达90%以上。结论:机器学习技术可揭示疾病症状和治疗方案之间的内在联系,有助于提高在线疾病咨询的效率,提供有针对性的备选治疗方案。

Objective To analyze and predict the tendency for online disease diagnosis and treatment plan using machine learning technique. Methods The data of cardiovascular disease-related 439 symptoms were crawled from The 39 Disease Encyclopedia Web. The keyword vector of symptoms was extracted using TF-IDF algorithm. A classification model was established with support vector machine, decision tree and neural network. The different tendencies of disease to Western medicine or combined Western and traditional Chinese medicine were predicted. Results Text analysis of online disease information could mine the characteristics of different diseases. Machine learning technique could effectively predict the tendency to online disease diagnosis and treatment plan with an accuracy 〉90%. Conclusion Machine learning technique can display the relationship between disease symptoms and treatment plan, improve the efficiency of online disease consultation, and provide the candidate treatment plan for different diseases.