Application of an Improved AEMK Algorithm Based on Autoencoder and K-means in Cluster Analysis of Physical Examination Data

摘要

The K-means algorithm is commonly used in the analysis of traditional health examination data. However, the K-means algorithm is unsatisfactory in the face of a large number of high-dimensional data. Therefore, the Auto-encoder and K-means have been studied, combined with the powerful Autoencoder. With the advantages of dimensionality reduction and K-means' ease of use and scalability, a novel algorithm that combines Autoencoder with K-means is proposed. Distance-based improvement of the initial value selection method can greatly improve the health examination. With the accuracy of data clustering analysis, the clustering coefficient of the improved algorithm has greatly improved, and it has great potential for application in health status analysis and disease prediction.