The Use of a Robust-Adaptive Self Organizing Map to Enhance the Prediction Performance of Clinical Datasets

Prediction in the medical field is a challenging problem and as a result many researchers have used different artificial intelligent methods including conventional Self Organizing Map (SOM) to achieve this task. SOM is a specialized clustering technique that has been used in a wide range of applications to solve different problems. Unfortunately, conventional SOM suffers from slow convergence and high steady-state error. The work presented in this paper is based on the recently proposed modified SOM technique introducing a Robust Adaptive learning approach to the SOM (RA-SOM). RA-SOM helps to overcome many of the current drawbacks of the conventional SOM and is able to outperform the SOM in obtaining the winner neuron in a lower learning process time. The efficient and outstanding performance achieved by applying RA-SOM in other research areas is the main driving force behind this work. To verify the improved performance of the RA-SOM, its performance is compared against the performance of other versions of the SOM algorithm, namely GF-SOM, PLSOM, and PLSOM2. The test results proved that the RASOM algorithm outperformed the conventional SOM and the other algorithms in terms of prediction time and accuracy. The test results also showed that RA-SOM maintained an efficient performance on the different datasets used, while the case of the other algorithms a more inconsistent performance was recorded, which means that their performance are data type-related.