The main objective of the proposed study is to integrate machine learning approach with healthcare databases in order to dig for useful information and extract patterns from larger datasets. In particular, the study will reveal insights from Kawasaki disease data sets that include class labels like gender, age at diagnosis, fever days along with features that describe treatment strategy and done in which field some complications occurred or whether the child under follow-up visits. Due to the dataset complexity, it was necessary to use different machine learning models such as Logistic Regression, Support Vector Machine (SVM), XG Boost and Naive Bayes for analyzing this data. The aim of the research was to obtain an optimized model for a highly accurate representation of data classification and retrieving patterns that may help health practitioners mainly across the world. The models were implemented in Python 3.9, using libraries like Scikit-learn to fine-tune the model performance measurements. Results show the accuracy scores of all Models: Logistic Regression-68% SVM-68%, XG Boost-52%, Naïve Bayes-48%. Machine learning provides a promising approach to contribute to addressing these gaps, which may help early diagnosis and management of the disease while supporting personalized treatment decisions with long-term repercussions on patient protection.