Search for: All records

We use evidence from a disruption of clinical documentation routines to propose a novel, predictive mechanism for routine dynamics based on path coherence. Path coherence refers to the continuity of situational attributes from one event to the next along a path, for example, a set of activities conducted by the same person has high actor coherence. Situational attributes include classic descriptors such as who, what, when, where, and why. To be recognized as a path, a minimal level of coherence is required, but path coherence can vary along a path. For example, in a medical clinic, typical paths flow from place to place (e.g., reception, waiting room, exam room) and involve different clinical staff (e.g., receptionist, nurse, physician). Using latent factor network models, we compare clinical documentation routines in five outpatient clinics before and after a technological disruption (an upgrade to the electronic health record system). We show that coherent paths are up to 14 times more likely to persist and up to 40 times more likely to form than less coherent paths. We use these findings to theorize about the role of path coherence in routine dynamics. Path coherence in narrative networks is like homophily in social networks, but with a completely different underlying mechanism. We discuss the implications of our findings for organizational path dependence, resilience, and inertia. Funding: This research was supported by the National Science Foundation [Grants SES-1734237 and BCS-2120530]. This research was also supported in part by the University of Rochester CTSA [Grant UL1 TR002001] from the National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health (NIH). 

The growing availability of digital trace data has generated unprecedented opportunities for analyzing, explaining, and predicting the dynamics of process change. While research on process organization studies theorizes about process and change, and research on process mining rigorously measures and models business processes, there has so far been limited research that measures and theorizes about process dynamics. This gap represents an opportunity for new information systems research. This research note lays the foundation for such an endeavor by demonstrating the use of process mining for diachronic analysis of process dynamics. We detail the definitions, assumptions, and mechanics of an approach that is based on representing processes as weighted, directed graphs. Using this representation, we offer a precise definition of process dynamics that focuses attention on describing and measuring changes in process structure over time. We analyze process structure over two years at four dermatology clinics. Our analysis reveals process changes that were invisible to the medical staff in the clinics. This approach offers empirical insights that are relevant to many theoretical perspectives on process dynamics.