Determination of Patient’s Hearing Sensitivity using Data Mining Techniques

The aim of this paper is to analyse and classify an audiometric dataset belongs to 200 patients using data mining techniques. The classification task consists of the classification the data of each patient as belonging to one out of four categories (conductive type, sensorineural hearing loss, mixed type or normal). These classifications show the diverse kinds of hearing misfortune relying upon which part of the hearing way is influenced. The used dataset provides sensitivity test values of the patient's hearing sense and these tests are mostly done by an audiologist using the audiometer experiments. The audiometric tests are used to define the patient's hearing threshold level at several frequencies. The specialists often try to locate the problem of hearing path where exactly, so this characterizes the hearing problem as having a place with one of the classified groups. There are several new data mining algorithms applied in this research to classify patient hearing system. This automated system will help medical specialists to make decision. Experimental results show that Artificial Immune Recognition System (AIRS), Pruned Tree J48 and Random Forest (RF) techniques have given very promising classification results (between 99-100% accuracy) for hearing loss data set. We have better accuracy results than other traditional machine learning methods for hearing dataset.