Applying data mining for the analysis of breast cancer data

Area of Science:

• Biomedical Informatics
• Machine Learning
• Computational Biology

Background:

• Data mining and artificial intelligence (AI) are crucial for extracting patterns from large datasets.
• Predictive modeling for breast cancer aids in early detection and treatment planning.

Purpose of the Study:

• To investigate the application of AI and data mining techniques for breast cancer prediction.
• To compare the performance of artificial neural networks, decision trees, logistic regression, and genetic algorithms.

Main Methods:

• Utilized a dataset of 699 breast cancer patients from the University of Wisconsin.
• Applied artificial neural network, decision tree, logistic regression, and genetic algorithm models.
• Employed tenfold cross-validation for model evaluation, focusing on accuracy and positive predictive value.

Main Results:

• Genetic algorithm model achieved the highest accuracy (0.9878), significantly outperforming others.
• Logistic regression and decision tree models showed moderate accuracy (0.9434).
• Artificial neural network model yielded an accuracy of 0.9502.

Conclusions:

• The genetic algorithm model demonstrates superior performance for breast cancer patient data classification.
• AI and data mining techniques, particularly genetic algorithms, offer accurate and comprehensible insights for breast cancer analysis.
• The study highlights the potential of advanced computational methods in improving breast cancer diagnostics.