The invention provides a prediction method and
prediction system of an irresponsive
gamma globulin Kawasaki disease. The method comprises the following steps that 21 original parameters of SVM
model building are collected; the modelling original parameters comprise gender, age, fever time during treatment, clinical classification, CRP detection value, WBC value, PLT value, Hb value, ALT value, AST value, ALB value,
gamma globulin using time and clinical diagnosed symptom indicators; the clinical diagnosed symptom indicators comprise conjunctival injection, erythra, cracked lips, a strawberry-like tongue, lymphadenectasis of a neck, hard and swollen hands and feet, digit peeling, crissum peeling and a red and swollen bacillus calmette-guerin scar;
discretization is conducted on the original parameters to obtain SVM characteristic values corresponding to the original parameters; the SVM characteristic values are regarded as base data, a SVM model is built, and through the SVM model, a complication with irresponsive
gamma globulin of the
Kawasaki disease is predicted. By means of the prediction method and
prediction system, early intervening treatment can be conducted on a patient,
recovery of damage to a coronary
artery is promoted, and the prediction method and
prediction system have important significance and value on diagnosis and treatment of the
Kawasaki disease in the future.