Diagnosis using clinical/pathological and molecular information.

Area of Science:

• Biomedical research
• Data integration
• Bioinformatics

Background:

• Integrating diverse patient data (molecular, clinical, pathological) is crucial for disease diagnosis and classification.
• Existing methods often analyze clinical and genomic data independently, potentially losing valuable information.
• Novel approaches are needed to effectively combine continuous gene expression data with categorical/ordinal clinical data.

Purpose of the Study:

• To introduce and evaluate the related metric scaling distance for integrating clinical/pathological and molecular data in biomedical research.
• To demonstrate the utility and advantages of this distance measure compared to other proximity methods.
• To assess its performance in clustering and discriminant analysis for disease classification.

Main Methods:

• The study presents the related metric scaling distance, a specialized metric for heterogeneous data integration.
• Comparison with alternative proximity measures for combining clinical and genetic information.
• Application of the distance in classical clustering and discriminant analysis, contrasted with specialized integration methods.
• Validation using simulated data and real-world datasets from heart disease and cancer studies.

Main Results:

• The related metric scaling distance effectively integrates diverse patient data, yielding competitive results.
• Demonstrated flexibility and availability of the distance measure across various datasets.
• Outperformed or showed comparable results to more complex, specialized integration techniques in certain analyses.

Conclusions:

• The related metric scaling distance is a valuable and effective tool for integrating molecular and clinical data in disease research.
• Its application can improve disease knowledge, classification, and capture information lost in independent analyses.
• The method offers a flexible and robust approach for analyzing complex biomedical datasets.