Process Mining a Healthcare Delivery System for Hetero-Functional Graph Analysis

Hetero-functional Graph Theory (HFGT) has been shown to provide analytical insights into the structure and behavior of complex engineering systems that cannot be achieved by other graph theoretic approaches. Hetero-functional graphs have been applied to many applications. Many of these are technological systems where the associated hetero-functional graphs are constructed from first principles founded in physics. In complex socio-technical systems, however, the hetero-functional graphs do not necessarily have physics-based first principles. Consequently, the difficulty with conducting HFG theoretic analyses is in constructing the graph in a systematic and time-efficient manner. Healthcare delivery is a probabilistic system heavily influenced by patient and provider decisions. Therefore, modeling healthcare delivery requires direct modeling from the data generated from utilizing the healthcare delivery system, such as claims-based data. Here, we provide a process mining approach for producing hetero-functional graphs of healthcare delivery systems from raw claims-based data, such as Medicare claims data. This process mining approach allows for the development of an automated and efficient means of producing hetero-functional graphs of the American healthcare delivery system and the analysis of the structure and behavior of these systems. Consequently, in this paper, we develop a healthcare delivery system process mining approach for hetero-functional graph analysis.