A novel Chinese herbal medicine clustering algorithm via artificial bee colony optimization

Area of Science:

• Computational intelligence
• Data mining
• Traditional Chinese Medicine (TCM) analysis

Background:

• Traditional Chinese Medicine (TCM) is globally recognized, increasing the need for robust TCM data analysis.
• Traditional clustering algorithms often rely on empirical values for selecting cluster centers, limiting accuracy.
• Automating cluster center selection is crucial for objective analysis of complex TCM datasets.

Purpose of the Study:

• To develop an improved artificial bee colony algorithm (IABC-DP) for accurate clustering of Chinese herbal medicines.
• To address the limitations of empirical cluster center selection in traditional algorithms.
• To enhance the analysis and discovery of composition rules in TCM prescriptions.

Main Methods:

• An enhanced artificial bee colony algorithm (IABC) with a novel neighbor nectar searching strategy was developed.
• The IABC algorithm was employed to optimize key parameters (cutoff distance, local density, minimum distance) for the Density Peak (DP) clustering algorithm.
• The IABC-DP algorithm was validated on UCI benchmark and TCM datasets, comparing performance against K-means, K-mediods, and DBSCAN.

Main Results:

• The IABC-DP algorithm demonstrated superior clustering accuracy and quality compared to classical algorithms on both benchmark and TCM datasets.
• Evaluation metrics including Silhouette Coefficient, Entropy, Purity, Precision, Recall, and F1-Measure confirmed the effectiveness of IABC-DP.
• The improved artificial bee colony algorithm significantly reduced iteration counts compared to traditional bee colony algorithms.

Conclusions:

• The IABC-DP algorithm provides an accurate and efficient method for clustering multi-dimensional Chinese herbal medicines.
• This approach advances the study of traditional Chinese prescription composition rules.
• The optimized clustering offers a reliable tool for TCM data mining and research.