Utilize Medical Text Data to Estimate Disease Types by Using Naïve Bayes and ANN Classifier

Abstract

The primary concept of the hospital isthe provision of health services to the community. In many cases,
the utilization of information technology to record all hospital activity data can improve hospitals'
quality services. However currently, the data is only stored in the database and used as history without
further use. Many experiences show that optimizing data usage can greatly assist doctors in making
decisions to minimize medical errors. For example, examination data that among others of anamnesis
(medical abstract), blood pressure, temperature, and other patient’s symptom data can be used to
classify the kind of disease. One of the challenges in medical data utilization is that these data consists
of various formats, structured, and unstructured as well. In this study, we address the medical
unstructured data format by using Natural Language Processing approach. The combination of its
representation results with the structured format data is then used as the dataset to build the model for
disease type prediction based on Naïve Bayes and Artificial Neural Network classifier. By using these
two algorithms, the results of the classification of the kind of disease. The performed experiments show
that the ANN model performs better with the best accuracy average of 89.29% compared to Naive
Bayes, which is 80.60 %.