Session 5: Tools and methodologies
4.3 Computational creativity support
The FRAM-based ontology for healthcare systems collects semantic descriptions of potential functions that an analyst may define when designing FRAM models for specific healthcare safety analysis problems. The FRAM-based ontology allows organizing such function descriptions in a knowledge base that can be used by the analyst to search information through semantics-based query functions. More interestingly, such formalized knowledge may suggest input parameters for FRAM models leveraging automatic reasoning methods, such as concepts/function models subsumption relations and similarity/dissimilarity metrics, as well as contextual or design related constraints. This task is accomplished by the Creativity Machine component of the safety imagination framework that implements methods of computational creativity, a subfield of Artificial Intelligence aiming at defining computational systems that create artifacts and ideas (Colton & Wiggins, 2012). Generally, computational creativity methods address the problem of thinking something new, e.g., a risk situation, by varying and/or combining one or more aspects of what already exists, e.g., old experiences of incidents or normal situations. For our application, we initially focus on the following methods: the transformation method, defined as the process that modifies the form of some particular features of an existing design; and the analogy method, defined as the process where specific aspects of the conceptual structure of one problem or domain are matched with and transferred to another problem or domain. In particular, we apply these definitions to the design of a FRAM function and of its couplings.
Given a collection of “ground-level” function semantic descriptions, a task of the safety analyst is to identify all the functions to include in a model, to define the aspects of each FRAM function and its couplings with the other functions. Indeed, the information from the healthcare personnel could be incomplete, and they could miss unusual or abnormal situations that are relevant to the FRAM analysis.
Thus, following the analogy method, support to specify aspects of a given FRAM function can be provided, for example, by showing to the analyst the aspects of similar functions. The description of the function “administer special analgesic” in the abdominal surgery model of Figure 2, with details like the choice/availability of the right analgesic as in the case study, could suggest aspects for “administer epidural anesthetic” function while analyzing labor for childbirth process.
As a FRAM coupling is automatically realized by identifying pair of aspects of different functions addressed by the same name, whenever two aspects refer to concepts that are in a subsumption relation in the FRAM-based healthcare ontology (i.e., they belong to the same taxonomy), the system may suggest the FRAM analyst a coupling between the two functions, or to abstract/further detail one of the two aspects/functions. The transformation method can be implemented by suggesting changes to function aspects or to couplings. In the example of Figure 2, “count instruments and materials before suturing” could be de-coupled from “suturing the wound” to suggest the analyst a situation that may occur that could be taken into account in the analysis, like that task is forgotten by distraction and not reported, as described in the case study.
5.
Conclusion
Foresight is the process of inferring new knowledge from pre-existing one. Enhancing the knowledge gathering process should be considered as a precondition to improve existing foresight methods. In this context we presented a novel framework with two objectives. The former is to increase engagement of sharp-end operators by means of a gamification approach and of the FRAM method. The latter is to use elicited knowledge and computational creativity methods to support safety analyst in thinking out of the box and in conceiving unimagined situations relevant to safety analysis.
The concrete example of application of this framework in the healthcare sector and a first positive feedback from safety analysts demonstrate, from one side, the need of novel more engaging approaches to collect expert knowledge and, from the other, the need to further increase the computational creativity support of our framework.
References
Borst, W. N. (1997). Construction of engineering ontologies for knowledge sharing and reuse. Universiteit Twente.
Coletti, A., De Nicola, A., & Villani, M. L. (2017). Enhancing creativity in risk assessment of complex sociotechnical systems. Lecture Notes in Computer Science (Vol. 10405 LNCS).
Colton, S., & Wiggins, G. A. (2012). Computational creativity: The final frontier? In Proc. of the 20th European conference on artificial intelligence (pp. 21–26). De Nicola, A., & Missikoff, M. (2016). A lightweight methodology for rapid ontology
engineering. Communications of the ACM, 59(3).
Gangemi, A., Guarino, N., Masolo, C., Oltramari, A., & Schneider, L. (2002). Sweetening ontologies with DOLCE. Lecture Notes in Computer Science, Vol. 2473, 223–233.
Gruber, T. R. (1993). A translation approach to portable ontology specifications. Knowledge Acquisition, 5(2), 199–220.
Hackos, J.T., Redish, J.C., 1998. Conducting the site visit - honing your interviewing skills. John Wiley & Sons, New York, NY.
Hernan, A.L., Giles, S.J., Fuller, J., Johnson, J.K., Walker, C., Dunbar, J.A., 2015. Patient and carer identified factors which contribute to safety incidents in primary care: a qualitative study. BMJ Qual. Saf. 24, 583–593.
Hollnagel, E., 2012. FRAM: The Functional Resonance Analysis Method: Modelling Complex Socio-technical Systems. Ashgate.
Niles, I., & Pease, A. (2001). Towards a Standard Upper Ontology. The 2nd International Conference on Formal Ontology in Information Systems, 2–9. Patton, M.Q., 2002. Qualitative research and evaluation methods, 3rd edition. ed. Sage
Publications, Thousand Oaks, CA.
Reeves, B., & Read, J. L. (2009). Total Engagement: How Games and Virtual Worlds Are Changing the Way People Work and Businesses Compete. Harvard Business Press.
Stanford. (2016). Protégé Ontology Management System,
http://protege.stanford.edu/about.php.
Woods, D.D., Cook, R.I., 2002. Nine Steps to Move Forward from Error. Cogn. Technol. Work 4, 137–144. doi:10.1007/s101110200012