Multi-level medical periodic patterns from human movement behaviors.

Area of Science:

• Data Mining
• Medical Informatics
• Geospatial Analysis

Background:

• Human movement behaviors captured by location-aware devices generate valuable spatio-temporal trajectory data.
• Existing methods for medical data mining often overlook spatio-temporal semantics and hierarchical pattern structures.

Purpose of the Study:

• To develop a novel medical data mining framework for generating multi-level medical periodic patterns from human movement trajectories.
• To effectively classify trajectories based on visits to medical centers and identify hierarchical medical behaviors.

Main Methods:

• Utilized spatio-temporal trajectory data from the Geolife dataset.
• Developed a framework to generate multi-level medical periodic patterns, incorporating spatio-temporal semantics.
• Applied the framework to distinguish between individuals who periodically visit medical centers and those who do not.

Main Results:

• The proposed framework successfully distinguished individuals who periodically visit medical centers from those who do not.
• Identified multi-level medical periodic patterns, revealing complex hierarchical medical behaviors.
• Demonstrated the feasibility and applicability of the framework using real-world trajectory data.

Conclusions:

• The framework offers a robust method for analyzing human movement data for medical insights.
• Enables automated personalized medical services, such as targeted health information delivery.
• Supports enhanced disease discovery and diagnosis through the identification of subtle medical patterns.