Making Bayesian Networks available for clinical decision support
Type: Bachelor CS
If you are interested please contact :
Evidencio (www.evidencio.com) is a company that hosts an online platform for medical prediction models and clinical decision support tools. The platform enables users to easily create an interactive interface for a model they’ve created, and utilize the platform to disseminate the developed model. Models can be used to calculate clinically relevant outcomes with the use of patient specific parameters for a large amount of publicly available prediction models. In addition, a validation module is available to assess the predictive performance for publicly accessible models. Medical professionals can utilize the content on the Evidencio platform to integrate clinical decision support tools directly into their clinical workflow. The majority of available models have been developed using multivariable regression analysis. Even though regression models have been deemed useful for clinical prediction models, they have several limitations regarding clinical utility. For instance, to estimate a prediction, info on all model parameters is required. In clinical practice, one or more required parameters are often missing in patient. Different modelling techniques, such as a Bayesian networks, enable to calculate predictions for a patient when info on one or more parameters is missing. As of today, models such as Bayesian networks cannot be added to the Evidencio platform as easily as regression models can be added. In addition, Bayesian networks provide much more useful information when visualized correctly. It would be greatly appreciated to expand the possibilities of creating models on the Evidencio platform that can be used to improve clinical decision making by means of, for instance, Bayesian networks.
The aim of Evidencio is that users of the platform should be able to create their own model and therefore should also be able to create their own visualization for the model. As there may be several obstacles to overcome this challenge, software should be developed for the Evidencio platform. The development and implementation of new software components is performed according to existing standards and guidelines. Before new components will be available for general use, the software components have to be specified, verified, and validated thoroughly on the development site of Evidencio. The development site is a separate IT infrastructure that is basically a copy of the platform available at www.evidencio.com. However, new software components that were created and tested by local development environments (i.e. software developers) will be verified and validated thoroughly on the development site. The development site is only accessible when working at the Evidencio office, located in Haaksbergen. To develop new components for the platform, experience with Laravel, PHP, and Vue is required. To fulfill this assignment, it is required to work at least one day a week at the office of Evidencio B.V. located in Haaksbergen. Several clinical Bayesian networks are already available and can be used to demonstrate the usefulness of the developed software.