A Comparative Study of Nine Machine Learning Techniques Used for the Prediction of Diseases

Danielle Azar, Rebecca Moussa, Georges Jreij

2018 · 13 citations · 0 references

Concepts

Abstract

We present a comparative study of nine state-of-the-art machine learning techniques for classification of diseases. These are: ID3, Deep Learning, Artificial Neural Networks, Naive Bayes, Logistic Regression, Partial Decision Trees, k-Nearest Neighbor, Classification via Clustering and Voting Feature Intervals. We test these techniques on eight datasets for the classification of different diseases. The data sets vary in their characteristics. We use two modeling techniques: cross-validation and boot-strapping. We assess the machine learning techniques using the following performance metrics: accuracy, precision and area under the ROC curve. Results show that they can be very beneficial over a wide range of diseases.