BRCA Variations Risk Assessment in Breast Cancers Using Different Artificial Intelligence Models.

Area of Science:

• Computational biology
• Oncology
• Artificial Intelligence

Background:

• Early cancer diagnosis significantly reduces mortality rates.
• Artificial intelligence (AI) simulates human brain functions for machine learning and decision-making.
• Integrating genetic variations with other risk factors can improve cancer risk assessment.

Purpose of the Study:

• To develop an AI system for breast cancer risk assessment by combining genetic variations and other risk factors.
• To evaluate the performance of AI models in predicting breast cancer risk based on multiple factors.

Main Methods:

• Analyzed data from 268 breast cancer patients, including 16 risk factors and genetic variants (BRCA1, BRCA2).
• Trained AI models using Mamdani's Fuzzy Inference Method and Feed-Forward Neural Network on MATLAB.
• Validated model performance through 16 tests on subjects new to the system.

Main Results:

• The neural network model achieved high success rates: 99.9% for training, 99.6% for validation, and 99.7% for testing.
• Fuzzy logic model demonstrated 95.5% accuracy.
• Both AI models provided similar predictive results, outperforming previous studies by incorporating a wider range of risk factors, including genetic data.

Conclusions:

• Developed AI models show significant promise for assessing breast cancer risk, particularly concerning BRCA variations.
• These AI systems offer a unique tool for personalized medicine, enhancing risk prediction accuracy.
• The study highlights the potential of AI in improving early detection and management of breast cancer.