OBJECTIVES: Decision models rely on a foundation of health states or events, along with their associated probabilities of transitioning between states over a specified time period, known as "transition probabilities." Modelers face challenges converting non-probabilistic data (counts, rates, etc.) from published sources into probabilities. Additionally, matching the cycle length of decision models with published probabilities, which may differ (e.g., annual vs. 6-month cycles), poses another common challenge. This tool aims to gather evidence on transition probability calculations from various sources for economic models, consolidating it into an open-source unified R-shiny platform.

METHODS: A web application was developed in R using the R-Shiny package. The tool was deployed using the docker container on Amazon Web Services (AWS) with Secure Sockets Layer (SSL) certificates, and Auth0 handled the user’s authentication. The input data uploaded on this platform will only last till the active session and will not be stored on any server. This tool assists modelers in converting information, including relative risks, odds ratios, and median or mean survival time, sourced from journals or government databases, into transition probabilities for state-transition models.

RESULTS: The web-based tool was deployed in open source and accessible by clicking on https://shubhrampandey.shinyapps.io/probabilityCalc/. With a simple click, modelers can instantly copy the calculated transition probabilities to the clipboard for seamless pasting into Excel or R models. Furthermore, a downloadable Word report provides a step-by-step breakdown of the calculation process from published evidence, serving reporting purposes.

CONCLUSIONS: Decision modelers encounter various challenges when incorporating transition probabilities from published data, including the reliance on comparative statistics (e.g., RRs or ORs) and the need to align data with the model's cycle length. This tool offers a solution, enabling modelers to populate their models with precise transition probabilities effortlessly, even without prior programming knowledge.