Classification of Childhood Diseases with Fever Using Fuzzy K-Nearest Neighbor Method

Fever or pyrexia is a condition when the body temperature rises above the average. This may occur due to viral or bacterial infection of the body. In addition, fever is the main symptom of diseases such as dengue fever, typhoid fever, diarrhea, gastroenteritis, measles, pneumonia, pharyngitis, and bronchitis. These diseases have similar symptoms, causing difficulty to distinguish them. In fact, the symptoms of diseases are usually recorded in a medical record document. Medical records can be categorized in order to ease diagnosis. The technique to categorize based on certain characteristics to several classes is called classification. Classification can categorize textual data which are first converted into numerical data so that the classification process can generate results. Fuzzy K-Nearest Neighbor is one classification technique that measures the distance between training and testing data, which then put them into a fuzzy set. This study developed a classification system for childhood diseases with fever using Fuzzy K-Nearest Neighbor based on textual medical record documents. The test results of the classification system showed an accuracy of 83.3% in the dengue fever and pneumonia data with a comparison of training and testing data of 80: 20, K value of 10, and M value of 2. Thus, it can be concluded that Fuzzy K-Nearest Neighbor classification system can be used as a solution to the classification of childhood diseases with fever.