@article{Torres-Sospedra2021a,
title = {Process Model Metrics for Quality Assessment of Computer-Interpretable Guidelines in PROforma},
author = {Joaquín Torres-Sospedra and Begoña Martínez-Salvador and Cristina Campos Sancho and Mar Marcos},
doi = {https://doi.org/10.3390/app11072922},
issn = {2076-3417},
year = {2021},
date = {2021-03-25},
journal = {Applied Sciences},
volume = {11},
number = {7},
pages = {2922},
abstract = {Background: Clinical Practice Guidelines (CPGs) include recommendations to optimize patient care and thus have the potential to improve the quality and outcomes of healthcare. To achieve this, CPG recommendations are usually formalized in terms of Computer-Interpretable Guideline (CIG) languages. However, a clear understanding of CIG models may prove complicated, due to the inherent complexity of CPGs and the specificities of CIG languages. Drawing a parallel with the Business Process Management (BPM) and the Software Engineering fields, understandability and modifiability of CIG models can be regarded as primary quality attributes, in order to facilitate their validation, as well as their adaptation to accommodate evolving clinical evidence, by modelers (typically teams made up of clinical and IT experts). This constitutes a novel approach in this area of CIG development, where understandability and modifiability aspects have not been considered to date. Objective: In this paper, we define a comprehensive set of process model metrics for CIGs described in the PROforma CIG language, with the main objective of providing tools for quality assessment of CIG models in this language. Methods: To this end, we first reinterpret a set of metrics from the BPM field in terms of PROforma and then we define new metrics to capture the singularities of PROforma models. Additionally, we report on a set of experiments to assess the relationship between the structural and logical properties of CIG models, as measured by the proposed metrics, and their understandability and modifiability from the point of view of modelers, both clinicians and IT staff. For the analysis of the experiment results, we perform statistical analysis based on a generalized linear mixed model with binary logistic regression. Results: Our contribution includes the definition of a comprehensive set of metrics that allow measuring model quality aspects of PROforma CIG models, the implementation of tools and algorithms to assess the metrics for PROforma models, and the empirical validation of the proposed metrics as quality indicators. Conclusions: In light of the results, we conclude that the proposed metrics can be of great value, as they capture the PROforma-specific features in addition to those inspired by the general-purpose BPM metrics in the literature. In particular, the newly defined metrics for PROforma prevail as statistically significant when the whole CIG model is considered, which means that they better characterize its complexity. Consequently, the proposed metrics can be used as quality indicators of the understandability, and thereby maintainability, of PROforma CIGs},
keywords = {Software metrics, software models},
pubstate = {published},
tppubtype = {article}
}