Using bi-dimensional representations to understand patterns in COVID-19 blood exam data

Area of Science:

• Medical diagnostics
• Data science in healthcare
• Hematology

Background:

• Blood tests are crucial for medical diagnosis.
• Disease manifestations, like COVID-19, can vary significantly.
• Multivariate blood test data presents analytical challenges.

Purpose of the Study:

• To develop a method for extracting insights from blood test data.
• To identify patterns related to infection prevalence using advanced analytics.
• To explore the hematological implications of systemic diseases like COVID-19.

Main Methods:

• Utilized Uniform Manifold Approximation and Projection (UMAP) for dimension reduction.
• Applied DBSCAN clustering to identify distinct data groups.
• Employed statistical approaches to analyze blood test parameters and infection prevalence.

Main Results:

• Identified distinct clusters indicating infection prevalence ranging from 2% to 37%.
• Demonstrated the method's capability to uncover complex data patterns.
• Highlighted significant hematological implications, particularly on white-cell fractions, with relevant statistical p-values (0.03-0.1).

Conclusions:

• The proposed UMAP and DBSCAN method effectively analyzes blood test data for disease patterns.
• COVID-19 exhibits systemic effects with significant hematological impacts.
• This analytical approach can be applied to various diseases for discovering new data patterns.