Bioinformatics Approach to Cancer Prediction using Quantum Clustering Algorithm for Behavioral Similarity in Gene...

Area of Science:

• Bioinformatics
• Quantum Computing
• Computational Biology

Background:

• Accurate classification of gene expression data is crucial for cancer diagnosis.
• Existing clustering algorithms face challenges with high-dimensional and noisy biological data.

Purpose of the Study:

• To introduce a Hybrid Quantum K-Means Clustering Algorithm with automatic cluster detection.
• To enhance the classification accuracy of cancerous and non-cancerous gene expression data.

Main Methods:

• Utilizes Quantum Multi-Feature Mapping for state encoding and Swap Test for quantum distance estimation.
• Employs Quantum Gradient-Based Optimization for dynamic cluster number identification by minimizing intra-cluster variance.
• Incorporates a probability-proportional distance strategy for initial centroid selection.

Main Results:

• Achieved superior clustering performance on breast cancer datasets compared to existing quantum K-Means.
• Demonstrated improved Silhouette Score (0.641 vs. 0.601), Calinski-Harabasz Index (766.57 vs. 617.65), and Davies-Bouldin Index (0.659 vs. 0.704).
• Showcased enhanced cluster compactness, separation, and accuracy despite slightly higher time complexity.

Conclusions:

• The Hybrid Quantum K-Means Clustering Algorithm offers significant improvements in accuracy and feasibility for gene expression data classification.
• The algorithm's resilience to quantum noise and efficiency with high-dimensional data make it promising for bioinformatics and cancer classification.
• Automatic cluster detection and optimized centroid selection contribute to its robust performance.