A Novel Bioinspired Algorithm for Mixed and Incomplete Breast Cancer Data Classification.

Area of Science:

• Medical Informatics
• Machine Learning
• Bioinformatics

Background:

• Accurate cancer pre-diagnosis is crucial for improving patient survival rates.
• Medical datasets often contain missing or mixed (numerical and categorical) values, posing challenges for existing classification algorithms.
• Few algorithms effectively handle datasets with both missing and mixed data types.

Purpose of the Study:

• To modify an existing classification algorithm to effectively handle datasets with missing and mixed values for improved cancer classification.
• To evaluate the performance of the modified algorithm against classical and bio-inspired methods.

Main Methods:

• Modification of the AISAC algorithm to create AISAC-MMD (Mixed and Missing Data).
• Training and testing AISAC-MMD on datasets with missing and mixed values.
• Comparative analysis with established algorithms like Nearest Neighbor, C4.5, Naïve Bayes, ALVOT, Naïve Associative Classifier, AIRS1, Immunos1, and CLONALG.

Main Results:

• The AISAC-MMD algorithm demonstrated significantly superior performance compared to classical and bio-inspired classification algorithms.
• Statistical analysis confirmed the enhanced classification accuracy of AISAC-MMD, particularly in breast cancer detection.
• The modified algorithm effectively addressed challenges posed by missing and mixed data types.

Conclusions:

• AISAC-MMD offers a robust solution for cancer classification, especially with complex, real-world medical data.
• The algorithm's ability to handle missing and mixed data significantly improves early cancer diagnosis accuracy.
• This advancement holds potential for enhancing patient survival through more effective early detection strategies.