| dc.description.abstract | This research has been dedicated to finding visualizations for the novel and rather abstract concept of the digital twin, which is often presented either as a flashy color figure or as some kind of model or results from data analysis. The idea of a twin is to summarize what defines a group of subjects based on their nearly identical features.
The idea of a twin is to summarize what defines a group of subjects based on their nearly identical features. In the health sector, we often consider patients and what constitutes good or optimal treatment for expected outcomes. We define different patient groups and also try to connect them with some form of decision support. In the case of the digital twin, the project aimed to illustrate how to identify a digital twin for a specific patient and assess the likelihood of a beneficial outcome.
In the field of arthroplasty, an attempt was made to create a digital twin by applying cluster analysis, event log analysis, and other methods to understand the potential clinical pathway a patient can take.
This project, through five iterations, generated several conceptual designs of digital twins. One design related a patient to a cluster where a digital twin could be found based on the similarity of the data. Another aimed to find a path that a patient is likely to follow based on their baseline data. The third model integrates both clusters and pathways to provide a more holistic picture of where a digital twin and a likely clinical pathway could be related.
A digital twin is a novel concept and its visualization relies on data analytical methods to extract the most significant features. The conceptual solutions of this project are general and allow for the inclusion of different kinds of results while maintaining a predefined way of navigating and connecting data. This helps users retain a consistent understanding of creating digital twins and interpreting the results. Cognitive walk-throughs were applied to assess the usability of these conceptual designs, which are in their final iteration as mid-fidelity prototypes.
Working with two different datasets, a quality data register and an intensive care unit dataset, has shown differences in the structure and intensity of information. The challenge is to create conceptual designs that can suit both datasets, but one feasible solution is to create different instantiations for the same conceptual design. Future work involves developing a fully functioning high-fidelity prototype that can operate on different databases while providing the same sense of the digital twin concept. | |
