Use of statistical analysis in the biomedical informatics literature

Area of Science:

• Biomedical Informatics
• Health Informatics
• Medical Informatics Research

Background:

• Statistics are fundamental to biomedical informatics.
• Understanding statistical method application is crucial for research quality.
• Previous analyses of statistical trends in informatics journals are limited.

Purpose of the Study:

• To analyze the utilization of statistical methods in biomedical informatics research.
• To compare statistical approaches used in two leading journals: JAMIA and IJMI.
• To identify trends in statistical application from 2000-2007.

Main Methods:

• A literature review of original investigation articles published between 2000 and 2007.
• Analysis of statistical methods categorized as: descriptive, elementary, multivariable, regression, machine learning, and other statistics.
• Comparative analysis of statistical method frequency between JAMIA and IJMI.

Main Results:

• Descriptive statistics were the most frequently employed method in both journals.
• Elementary statistics (t-tests, chi-squared, Wilcoxon) were more prevalent in JAMIA.
• Machine learning methods showed similar usage rates across both journals.
• Diagnostic statistics (sensitivity, specificity, precision, recall) were more common in JAMIA.

Conclusions:

• Biomedical informatics research extensively utilizes a range of statistical methods.
• Proficiency in descriptive and elementary statistics is essential for biomedical informatics scientists.
• Journal-specific trends indicate varying preferences for advanced statistical techniques.