Process-oriented Iterative Multiple Alignment for Medical Process Mining

Area of Science:

• Process mining
• Data visualization
• Bioinformatics algorithms

Background:

• Trace alignment, adapted from biological sequence alignment, visualizes and analyzes workflow data.
• Current trace alignment methods use progressive guide-trees, resulting in O(N^2 L^2) time complexity and interpretation errors.
• Existing techniques are computationally intensive and sensitive to guide-tree metrics, hindering analysis of large datasets.

Purpose of the Study:

• To introduce process-oriented iterative multiple alignment (PIMA) as a novel technique for workflow data analysis.
• To address the limitations of existing trace alignment algorithms, including speed and accuracy.
• To demonstrate PIMA's effectiveness in improving sum-of-pairs scores and computational efficiency.

Main Methods:

• Developed process-oriented iterative multiple alignment (PIMA) with specialized optimizations for workflow data.
• Implemented iterative alignment for enhanced data representation and insight extraction.
• Evaluated PIMA's performance against existing trace alignment algorithms using sum-of-pairs score and time complexity analysis.

Main Results:

• PIMA achieves a sum-of-pairs score superior to existing trace alignment algorithms.
• PIMA demonstrates a significantly improved time complexity of O(NL^2).
• The framework is flexible and effectively applied to medical workflow data analysis.

Conclusions:

• PIMA offers a more efficient and accurate approach to trace alignment for process mining.
• Iterative alignment using PIMA facilitates better data representation and insight extraction from workflow data.
• PIMA enhances the analysis and visualization of complex datasets, particularly in medical contexts.