Topic modeling for systematic review of visual analytics in incomplete longitudinal behavioral trial data

Area of Science:

• Biomedical Informatics
• Data Visualization
• Behavioral Science

Background:

• Longitudinal observational and randomized controlled trials (RCT) are crucial in biomedical behavioral studies and smart health.
• These trials generate complex data: high-dimensional, correlated, and with missing values, posing analytical challenges.
• Visual analytics methods for such data are currently underdeveloped.

Purpose of the Study:

• To systematically review visual analytic methods for longitudinal behavioral trials over 28 years.
• To compare these methods with MIFuzzy, a soft computing tool for incomplete longitudinal data.
• To identify trends and gaps in visual analytics for behavioral trial data.

Main Methods:

• Developed a longitudinal topic model for systematic review of visual analytic methods from IEEE VIS conference.
• Compared identified methods with MIFuzzy for pattern recognition, validation, and visualization of incomplete longitudinal data.

Main Results:

• Longitudinal topic modeling revealed distinct trend patterns in visual analytics development for behavioral trials.
• A significant gap exists between current robust visual analytic methods and practical algorithms for longitudinal behavioral trial data.
• MIFuzzy demonstrates potential for handling incomplete longitudinal data.

Conclusions:

• There is a substantial need for advanced visual analytic methods and algorithms tailored to complex longitudinal behavioral trial data.
• Future research should focus on bridging the identified gap to enhance smart health systems and behavioral studies.
• MIFuzzy offers a promising approach for analyzing incomplete longitudinal data in behavioral research.