Feature Engineering and Process Mining to Enable Hazard Detection in Health Information Technology

Area of Science:

• Health Information Technology (HIT)
• Process Mining
• Clinical Workflow Analysis

Background:

• Health information technology (HIT) systems require enhanced reliability.
• Clinical order transactions present complex workflows that are difficult to analyze.
• Lack of well-defined event logs hinders hazard detection in corporate data warehouses.

Purpose of the Study:

• To improve the reliability of health information technology (HIT) systems.
• To detect plausible hazards within clinical order transactions.
• To simplify the analysis of complex clinical order processes.

Main Methods:

• Identifying relevant timestamped data fields to generate raw event sequences from transaction data.
• Adopting OASIS Human Task standard state transitions to map event sequences.
• Utilizing process mining techniques for workflow analysis.

Main Results:

• A novel approach to identify potential HIT hazards in clinical order transactions.
• Methodology for mapping and simplifying complex clinical radiology order lifecycles.
• Demonstrated potential for investigating HIT system improvements using process mining.

Conclusions:

• The proposed approach offers a method for enhancing HIT system reliability.
• Process mining can effectively identify hazards and areas for improvement in clinical workflows.
• The methodology has potential for future applications in healthcare system analysis.