Abstract—Machine learning algorithms have been extensively used in various areas, especially for diagnosing medical conditions such as cardiovascular disease, cancer, diabetes etc. Most of the researches estimate the individual classification measures for the particular algorithm implemented on a given dataset or combine two algorithms, from possibly different machine learning groups, in different phases of data processing. This paper shows that, in case of a concurrent implementation of two (or more) classification algorithms, the classification quality can be significantly improved. The case study is built on the Support Vector Machine (SVM) and the Naïve Bayes classifier (NBC) in detection of diabetic or pre-diabetic condition. The proposed hybrid system improves the overall computer-based accuracy for diabetes classification to the value of around 98%, and reduces the false negative diagnosis to the value of 0.7 %. The results show that SVM over performs NBC in diabetes detection, while joint implementation over performs both classifiers individually. The proposed system/approach can be adapted for constructing the support tools in medical diagnostics.

Index Terms—algorithms, diabetes, machine learning, support vector machine, naïve Bayes

Cite: Zhilbert Tafa, "Concurrent Implementation of Supervised Learning Algorithms in Disease Detection," Vol. 7, No. 2, pp. 124-128, May, 2016. doi: 10.12720/jait.7.2.124-128