On classification capability of neural networks: a case study with otoneurological data

Area of Science:

• Neuroscience
• Artificial Intelligence
• Machine Learning

Background:

• Accurate differentiation of otoneurological diseases is clinically significant.
• Machine learning methods offer potential for improving diagnostic accuracy.

Purpose of the Study:

• To evaluate the efficacy of multilayer perceptron and Kohonen neural networks in classifying challenging otoneurological diseases.
• To identify key factors influencing the performance of these neural networks in medical diagnostics.

Main Methods:

• Utilized multilayer perceptron neural networks for otoneurological disease classification.
• Employed Kohonen neural networks for otoneurological disease classification.
• Compared the performance of both neural network architectures.

Main Results:

• Both neural networks demonstrated efficiency in distinguishing between otoneurological diseases.
• Performance was contingent upon a uniform distribution of the learning dataset and an adequate number of learning cases.
• Kohonen neural networks exhibited superior performance when dealing with multiple input variables compared to perceptron networks.

Conclusions:

• Multilayer perceptron and Kohonen neural networks are viable tools for otoneurological disease recognition.
• Adherence to principles of uniform data distribution and sufficient sample size is critical for successful neural network application in this domain.
• Kohonen networks show promise for complex otoneurological disease classification tasks with numerous variables.