基于hLDA层次主题模型的患者投诉分类框架研究

Classification framework for patient complaints based on hierarchical latent Dirichlet allocation topic model

目的:使用机器学习方法构建患者投诉分类框架,以更好地评估医疗服务的质量。方法:使用hierarchical latent Dirichlet Allocation(hLDA)层次主题模型挖掘患者投诉数据的主题,再归纳出一个分类框架,并对患者投诉的主题分布进行可视化分析。结果:构建了一个基于患者投诉的分类框架,包括5个大类7个小类。可视化分析结果表明,患者投诉主要集中在临床服务类、环境类、管理类,共占88.37%。结论:通过对患者投诉数据进行挖掘,构建患者投诉分类框架,有利于更加科学合理地评估医疗服务质量。

Objective To improve the assessment of medical service by establishing the classification framework for patient complaints using machine learning.Methods The data of patient complaints were mined and a classification framework for patient complaints was established according to the hierarchical latent Dirichlet allocation(hLDA)topic model.The distribution of topics on patient complaints was analyzed by visual analysis.Results The classification framework for patient complaints which is consisted of 5 categories and 7 subcategories was successfully established.Visual analysis showed that the patient complaints were focused on clinical service,environment and management,accounting for 88.37%.Conclusion Establishment of the classification framework for patient complaints by mining their data can scientifically and rationally assess the medical service for patients.