摘要
甲状腺疾病检查的指标数量往往是根据医生经验的不同而变化的,由于对甲状腺疾病认识和早期诊断的缺乏,甲状腺疾病在很多不发达国家还是很严重的问题。在我国医疗水平不够完备地区依然缺乏甲状腺疾病经验丰富的医生。由于检查的指标数量以及周期性,患者也会承担比较大的经济和时间成本。介绍一种解决此问题的方法,该方法的第一个模块是通过特征选择处理甲状腺疾病的数据集,达到降低维度选取出部分特征子集。第二个模块是使用选择出来的特征子集训练分类模型。筛选出具有3-6个特征的数据集,使用其进行训练的分类模型准确率分别为97.30%、99.05%、99.07%和99.06%。而没有处理的原数据集所训练的分类模型准确率为98.13%。此方法在以后可以被研究用于其他疾病的诊断和预测,也可以重点研究如何对筛选出来的特征子集进行处理从而获得更高的准确度。
The number of indicator which doctors used to diagnosis often varies according the doctor’s experience.Thyroid disease is still a serious problem in many underdeveloped countries and areas due to the lack of awareness and early diagnosis of thyroid disease.There is still a shortage of experienced doctors in areas with inadequate medical level in China.Due to the large number of indicators and periodicity of the examination,patients will also bear a relatively large economic and time cost.This research introduces a method to solve this problem.The first module of this method is to deal with the data set of thyroid disease through feature selection,so as to reduce the dimension and select some feature subsets.The second module is to use a selected subset of features to train the classification model.In this research,the data set with 3-6 features were selected out and the accuracy of the classification model arrive at 97.3%,99.05&,99.07%and 99.06%re spectively.The accuracy of the classification model trained with original data set was 98.13%.This research can be used for the diagnosis and prediction of other diseases in the future and it can also focus on how to process the selected feature subset to obtain higher accuracy.
作者
肖钦文
XIAO Qin-wen(College of Computer Science,Sichuan University,Chengdu 610065)
出处
《现代计算机》
2020年第6期23-26,共4页
Modern Computer
关键词
甲状腺
降维
特征选择
诊断
Thyroid
Dimension Reduction
Feature Selection
AI Diagnosis
作者简介
肖钦文(1995-),男,四川成都人,硕士研究生,研究方向为人工智能、数字医疗。
