Data Mining of PubMed

Data Mining of PubMed mines PubMed entries to quantify keyword prevalence, temporal trends, and correlations among biomedical topics for literature-scale analysis.

Key Features:

• Keyword-Based Querying: Executes complex keyword combinations with temporal (year) filters to select PubMed entries.
• Trend Analysis: Calculates the proportion of PubMed entries matching keywords across years to detect shifts in topics such as Cancer prevalence.
• Comparative Analysis: Compares prevalence among conditions including Cancer, AIDS, and Malaria across time.
• Correlation Testing: Computes correlations between keywords to explore associations (for example, between Cancer and "DNA", "Computational", or "Mathematical").
• Organ-Specific Queries: Performs organ-related queries to analyze variation within cancer research.
• Visualization and Statistical Analysis: Produces visualizations and statistical summaries from query results.

Scientific Applications:

• Trend Identification: Identify emerging trends and shifts in scientific focus over time within PubMed literature.
• Research Impact Analysis: Assess the influence of technological and methodological advances, including computational biology, on fields such as oncology by examining keyword correlations.
• Resource Allocation Insights: Examine relationships between funding-related queries and research output across biomedical topics.

Methodology:

Relies on presence or absence of keywords within PubMed entries; constructs and executes keyword-based queries using Matlab; analyzes resulting datasets to generate visualizations and statistical summaries.