Breast cancer diagnosis using the fast learning network algorithm

Area of Science:

• Medical Informatics
• Machine Learning
• Computational Biology

Background:

• Machine learning (ML) and data mining show promise for breast cancer (BC) diagnosis.
• Existing ML approaches for BC diagnosis often lack rigorous statistical evaluation or use insufficient metrics.
• The Fast Learning Network (FLN) is an effective ML algorithm not yet applied to BC diagnosis.

Purpose of the Study:

• To propose and evaluate the Fast Learning Network (FLN) algorithm for enhanced breast cancer (BC) diagnosis.
• To assess FLN's capability in eliminating overfitting and handling both binary and multiclass classification problems.
• To compare FLN's performance against established classification methods in BC datasets.

Main Methods:

• The study implemented the Fast Learning Network (FLN) algorithm.
• Performance evaluation was conducted using two established breast cancer datasets: Wisconsin Breast Cancer Database (WBCD) and Wisconsin Diagnostic Breast Cancer (WDBC).
• Key performance metrics included accuracy, precision, recall, F-measure, G-mean, MCC, and specificity.

Main Results:

• The FLN algorithm achieved high performance on the WBCD dataset, with an average accuracy of 98.37% and other metrics exceeding 95%.
• On the WDBC dataset, FLN demonstrated strong results, achieving an average accuracy of 96.88% and high scores across all evaluated metrics.
• FLN effectively addressed overfitting and classification challenges, showcasing its robustness.

Conclusions:

• The Fast Learning Network (FLN) algorithm is a reliable and highly accurate classifier for breast cancer (BC) diagnosis.
• FLN's performance suggests its potential utility in other healthcare applications requiring robust data classification.
• The study highlights FLN as a significant advancement in the application of machine learning for medical diagnostics.