Privacy-Preserving Process Mining in Healthcare.

Area of Science:

• Health Informatics
• Process Mining
• Data Privacy

Background:

• Process mining offers valuable insights into healthcare operations but raises significant data privacy concerns due to sensitive patient information.
• Existing data mining privacy techniques are often unsuitable for anonymizing healthcare process data without adaptation.
• Strict privacy regulations can reduce the utility of healthcare data for process mining analyses.

Purpose of the Study:

• To analyze data privacy and utility requirements specific to healthcare process data.
• To assess the suitability of existing privacy-preserving data transformation methods for anonymizing healthcare event logs.
• To propose a framework for privacy-preserving process mining in healthcare settings.

Main Methods:

• Evaluation of privacy-preserving data transformation techniques from the data mining community.
• Analysis of the impact of anonymization methods on process mining results using three public healthcare event logs.
• Development of a conceptual framework for privacy-preserving process mining.

Main Results:

• Demonstrated how specific anonymization methods affect key process mining outcomes.
• Identified limitations of direct application of data mining privacy techniques to process data.
• Highlighted the need for tailored approaches to balance privacy and data utility in healthcare process mining.

Conclusions:

• A framework for privacy-preserving process mining is proposed to support healthcare analyses.
• Advocates for recording privacy metadata to track data transformations.
• Emphasizes the importance of adapting privacy techniques for effective and secure healthcare process mining.