Utilising Artificial Intelligence for Disease Classification and Prediction

Abstract

The main objective of this research is to investigate the role of artificial intelligence in disease classification and prediction. A brief review of the techniques, algorithms, tools and terminologies that were used in this work has been conducted. Artificial Neural Networks (ANNs) are reviewed to nominate the suitable type for this work.

In this work, a real medical data set has been used. The data set includes 14 attributes, of which 13 independent diagnosis variables and one categorical dependent variable, which is the type of heart disease.

To classify heart disease, a classification model is developed by using TensorFlow in Python. It is found that the classification model is 87% accurate in classifying heart disease. The challenges to implementing this model are explained, such as the data pre-processing, which means that the medical data cannot be used directly as some of them are categorical data that requires encoding before it can be used for the model development procedures.

It is concluded that data sets cannot be directly used after the acquisition because, for example, the data sets may include missing data and faulty readings, and these represent big challenges for the real-time processing and presentation requirements. It is also found that variables of different types, such as logical variables and categorical information, require encoding before using them to build prediction or classification models.