A MATLAB toolbox for classification and visualization of heterogenous multi-scale human data using the Disease State...

Area of Science:

• Biomedical Informatics
• Computational Biology
• Data Science

Background:

• Increasing volumes of human data necessitate advanced tools for information extraction.
• Accurate classification and exploration of multivariate patient data are critical in medical research.

Purpose of the Study:

• To present a MATLAB implementation for classifying and visually exploring multivariate patient data.
• To introduce the Disease State Index (DSI) and Disease State Fingerprint (DSF) methods for data analysis.
• To provide a robust and flexible tool for the biomedical research community.

Main Methods:

• Utilizes the Disease State Index (DSI) to measure data fit to predefined classes (e.g., controls vs. positives).
• Employs the Disease State Fingerprint (DSF) method for combining and visualizing DSI values in a tree-like structure.
• Develops a MATLAB toolbox to facilitate the application of these methods, robust to missing data.

Main Results:

• Demonstrates the classification and visualization capabilities of the DSI and DSF methods.
• Presents a functional MATLAB toolbox for researchers to implement the described methods.
• Illustrates the versatility of the toolbox through various personal health data classification examples.

Conclusions:

• The developed MATLAB toolbox offers a powerful and flexible solution for analyzing multivariate patient data.
• The DSI and DSF methods provide valuable insights into variable relevance for classification tasks.
• The implementation enhances the ability of biomedical researchers to extract meaningful information from complex datasets.