Automatic breast cancer diagnosis based on hybrid dimensionality reduction technique and ensemble classification

Area of Science:

• Computational Biology
• Machine Learning
• Medical Informatics

Background:

• High-dimensional data in medical diagnosis, such as breast cancer, often leads to overfitting and reduced efficiency.
• Effective feature selection is crucial for improving prediction accuracy and reducing decision time in large-scale datasets.
• Ensemble classifiers enhance classification model performance by integrating multiple individual models.

Purpose of the Study:

• To propose an advanced ensemble classifier algorithm for accurate breast cancer diagnosis.
• To enhance classification performance by integrating evolutionary computation with neural networks.
• To address challenges posed by high-dimensional data through hybrid dimensionality reduction.

Main Methods:

• Developed an ensemble classifier algorithm utilizing a multilayer perceptron neural network.
• Employed an evolutionary approach to optimize neural network parameters, including hidden layers, neurons, and weights.
• Implemented a hybrid dimensionality reduction technique combining Principal Component Analysis (PCA) and Information Gain.

Main Results:

• Evaluated the algorithm's effectiveness using the Wisconsin breast cancer database.
• Achieved an average accuracy improvement of 17% compared to existing state-of-the-art methods.
• Demonstrated superior performance in breast cancer classification tasks.

Conclusions:

• The proposed algorithm shows significant potential as an intelligent medical assistant system for breast cancer diagnosis.
• The integration of ensemble methods, evolutionary optimization, and hybrid dimensionality reduction enhances diagnostic accuracy.
• The findings support the use of advanced machine learning techniques in clinical decision support systems.