Disease danger factor extracting method based on improved K-means clustering

Abstract

The invention discloses a disease danger factor extracting method based on improved K-means clustering. The method comprises the following steps of firstly constructing a user information matrix and alabel vector according to a user investigation questionnaire of a disease; secondly, standardizing the user information matrix; then performing characteristic selection on the standardized user information matrix, calculating an intersection by means of methods of chi-square testing and eliminating a least variance, obtaining a related characteristic and constructing a problem characteristic datamatrix; afterwards, according to an improved Canopy algorithm and a K-means algorithm, performing clustering analysis on the characteristic attribute in a problem characteristic data matrix, and obtaining different class clusters; and finally, performing correlation coefficient analysis on each class cluster, selecting the characteristic with highest correlation coefficient as the representativecharacteristic of the class cluster, and adding the characteristic into a danger factor set. Relative to a frequency theory method in the danger factor extracting method in existing medicine field, the method is advantageous in that the disease danger factor can more efficiently and accurately extracted.

Description

technical field [0001] The invention relates to the fields of big data technology and medicine, in particular to a disease risk factor extraction method based on improved K-means clustering. Background technique [0002] Gastroesophageal reflux disease refers to a disease in which gastric reflux flows back into the esophagus, causing discomfort symptoms and complications. As a common clinical disease of the digestive system, it generally exists in various Asian and Western countries, and its incidence is increasing year by year. high trend. According to research, gastroesophageal reflux disease is related to various factors such as personal life, eating habits, and mental status, and the condition is prone to change. Therefore, exploring the risk factors of gastroesophageal reflux disease through big data technology is of great significance for the treatment and prevention of the disease. [0003] At present, for the risk factors of gastroesophageal reflux and other diseas...

AI Technical Summary

Problems solved by technology

[0006] In the current clustering method, there are two shortcomings in the traditional K-means algorithm: 1) the selection of the initial point, the selection of the initial point, there are many solutions, but there is still an inevitable problem of randomization; 2) Determination of the number of clusters. Before a deep understanding of the data, it is undoubtedly a big problem to select the number of clusters. Most of the existing methods are given based on the scoring mechanism, but once this method involves big data , there will be a large amount of calculation

The Canopy algorithm has the following disadvantages: 1) The clustering accuracy is low, and the algorithm forms several canopy that intersect with each other, and there is a large error; 2) There is a randomization part, that is, each time the center point of the canopy is selected randomly 3) It is necessary to manually determine the distance thresholds T1 and T2, and it is difficult to determine the distance

Images