Local and Distributed Machine Learning for Inter-hospital Data Utilization: An Application for TAVI Outcome Prediction

Background: Machine learning models have been developed for numerous medical prognostic purposes. These models are commonly developed using data from single centers or regional registries. Including data from multiple centers improves robustness and accuracy of prognostic models. However, data sharing between multiple centers is complex, mainly because of regulations and patient privacy issues.Objective: We aim to overcome data sharing impediments by using distributed ML and local learning followed by model integration. We applied these techniques to develop 1-year TAVI mortality estimation models with data from two centers without sharing any data.Methods: A distributed ML technique and local learning followed by model integration was used to develop models to predict 1-year mortality after TAVI. We included two populations with 1,160 (Center A) and 631 (Center B) patients. Five traditional ML algorithms were implemented. The results were compared to models created individually on each center.Results: The combined learning techniques outperformed the mono-center models. For center A, the combined local XGBoost achieved an AUC of 0.67 (compared to a mono-center AUC of 0.65) and, for center B, a distributed neural network achieved an AUC of 0.68 (compared to a mono-center AUC of 0.64).Conclusion: This study shows that distributed ML and combined local models techniques, can overcome data sharing limitations and result in more accurate models for TAVI mortality estimation. We have shown improved prognostic accuracy for both centers and can also be used as an alternative to overcome the problem of limited amounts of data when creating prognostic models.

BIM Information Standard Framework for Model Integration and Utilization Based on openBIM

The use of Building Information Modeling (BIM) is becoming more common in the construction process, and the demand for the integrated use of BIM data is expected to continue increasing. BIM is a technology that can maximize various efficiencies by sharing facility information in an integrated manner; however, the information generated in the life cycle of the construction industry is broad, diverse, and complex. Accordingly, because BIM information is difficult to share and is often duplicated, it is not easy to obtain the effects of various BIMs. The solution to this is to create a systematic standard so that one item of information can be used in various ways, and everyone shares and uses it together. To this end, various standards (guides, classification systems, information standards, etc.) are being created; however, the interrelationships between standard elements are complex, so there is confusion and overlap between standards. This paper proposes an information standard framework for BIM to identify a systematic standard and method to effectively develop various guidelines for the standard.

CREATION OF DIGITAL TWINS - KEY CHARACTERISTICS OF PHYSICAL TO VIRTUAL TWINNING IN MECHATRONIC PRODUCT DEVELOPMENT

Due to the falling costs of computational resources and the increasing potential of data acquisition, interest in digital twins, a virtual copy of the physical original, and their industrial application is increasing. Nevertheless, there is limited published work on how to support the process of physical to virtual twinning and what its key aspects are. The aim of this study is to present insights with regards to physical to virtual twinning gained from modelling projects in mechatronic product development. We conducted a survey and in-depth interviews with members of modelling projects. In the surveys and interviews we identified how physical products and virtual models were linked, which virtual models were used and which general challenges and key aspects are considered important by the project members. Our findings show that the key characteristics that pose challenges to modelling regarding physical to virtual twinning are model granularity, model validation, and model integration and interconnectivity.