STARS: Sharing Tools and Artifacts for Reproducible Simulation
Lead Research Organisation:
UNIVERSITY OF EXETER
Department Name: Health and Community Sciences
Abstract
Background:
Simulation models are computational tools that use detailed logic, data, and computer code to provide a quantitative way for researchers to make predictions about drug effectiveness, and health services operational flow. These models are used extensively in health and medical research to assess the effects of changes to patient care and to manage and understand pandemics like Covid-19. The most common approach used in these studies is called discrete-event simulation.
Challenges:
Very few published studies using discrete-event simulation meet the scientific standard for being open to (re)use, and scrutiny by others. In contrast, fields outside of healthcare, such as Ecology, have seen growth in model sharing. This means that healthcare results are more difficult to fully check or reproduce, and models are not tested for mistakes. Even when a model is shared with a scientific paper, there are considerable challenges in installing specialist simulation software, researcher concerns about intellectual property, time/effort needed to do the sharing, and how long a model remains available.
Simulation models are computational tools that use detailed logic, data, and computer code to provide a quantitative way for researchers to make predictions about drug effectiveness, and health services operational flow. These models are used extensively in health and medical research to assess the effects of changes to patient care and to manage and understand pandemics like Covid-19. The most common approach used in these studies is called discrete-event simulation.
Challenges:
Very few published studies using discrete-event simulation meet the scientific standard for being open to (re)use, and scrutiny by others. In contrast, fields outside of healthcare, such as Ecology, have seen growth in model sharing. This means that healthcare results are more difficult to fully check or reproduce, and models are not tested for mistakes. Even when a model is shared with a scientific paper, there are considerable challenges in installing specialist simulation software, researcher concerns about intellectual property, time/effort needed to do the sharing, and how long a model remains available.
