The University of Johannesburg (UJ) has developed a hybrid expert system – a smart system designed to augment and improve a medical specialist’s capabilities and accuracy of diagnosis – by combining various machine learning and regression methods, UJ said on Tuesday.
The research, published in the IEEE Xplore journal in 2019, combined theoretical and experimental physics and computer science to develop and train a new machine learning technique that predicts and diagnoses diseases such as lung cancer, tuberculosis, cardiovascular diseases and malaria.
In a case study to diagnose autism spectrum disorder cases, the proposed method achieved a prediction accuracy of the ensemble at above 96% with reduced variance, which is much better than those reported in the literature, said UJ Institute of Intelligent Systems and lead author Professor Qing-Guo Wang.
By fusing tree-based machine learning with a random order, the scientists believe the symptoms of an ailment from a patient serve as the input vector to diagnose the ailment using an artificial intelligence (AI) model.
“There is an urgent need for the development of easily implemented, automatic and effective screening methods. This will help health professionals and inform individuals whether or not they should pursue a formal clinical diagnosis,” the UJ team said.
The research used a suite of artificial neural networks (ANN) that were designed and trained to acquire knowledge about the task at hand. Based on this, machine learning pointed to a new forest-building method to support the process, said Wang.
The research was a collaboration with machine learning engineer Adeola Ogunleye and UJ Vice-Chancellor and Principal Professor Tshilidzi Marwala, said Wang.
“We combined ‘Decision Trees’ and regression methods, which are usually in two different branches of machine learning, to take advantage of each. A number of intelligent systems integrate two or more of the AI techniques – including supervised machine learning techniques support-vector machines and K-nearest neighbour – with a fuzzy logic system to form a hybrid expert system.”
The research, he said, was a milestone for randomisation to be introduced at tree growth and forest creation.
“The local prediction accuracies on the leaves are used to select a subset of the test data for actual predictions. The ensemble combines trees and thereby gives a better performance than the individually best-performing tree,” said Wang.
The UJ team has started a collaborative project with the Charlotte Maxeke Academic Hospital for AI-based diagnosis of breast cancer. UJ is also considering expanding the scope of this kind of AI work through significant funding to address more common diseases in the country.