Kawasaki Disease and Fever Diagnosis System Based on Incremental Local Discriminant Subspace Embedding

Abstract

The present invention provides a Kawasaki disease and fever diagnosis system based on incremental local discriminant subspace embedding, including a training subsystem and a diagnosis subsystem, wherein: the task of the training subsystem is to learn from the historical medical records of children with Kawasaki disease and common fever Generate clinical performance and detection physiological data for training, and then analyze local structure information to find the optimal local discriminant subspace, so as to establish Kawasaki disease and fever feature models in it; in addition, when new patient records are acquired, the training The subsystem updates the model in the form of incremental learning; the task of the diagnostic subsystem is to observe the clinical manifestations and detect physiological data of the patient to be diagnosed, and judge whether he has Kawasaki disease. The invention enables children suffering from Kawasaki disease to be treated in time, reduces the probability of developing coronary artery dilatation or aneurysm, and weakens the damage to the heart vessels of patients.

Description

technical field [0001] The present invention relates to a pediatric disease diagnosis system in the field of big data medical treatment, in particular to a Kawasaki disease and fever diagnosis system based on incremental local discriminant subspace embedding, which can be trained online, and can be used as a primary diagnostic tool for pediatricians. Diagnostics provide decision support. Background technique [0002] Kawasaki disease is a vasculitic syndrome, also known as mucocutaneous lymph node syndrome in children. It is a pediatric disease named after Dr. Tomisaku Kawasaki, Japan. Its clinical manifestations are fever, rash, swollen non-purulent lymph nodes in the neck, hyperemia of the conjunctiva of the eyes, diffuse hyperemia of the oral mucosa, bayberry tongue, erythema of the palms and soles, and hard edema of the hands and feet. Kawasaki disease occurs in children under the age of five, and the probability of boys getting it is about 1.5 times that of girls. The ...

AI Technical Summary

Problems solved by technology

However, this method has two disadvantages: 1) the classification uses linear discriminant analysis, which belongs to the linear global feature extraction method, and cannot fully mine the nonlinear local information hidden in the data; 2) the training method used requires batch Solving a generalized eigenvalue problem with data requires a large amount of calculation and is not easy to update online

Images