Presentation (CTI)

OBJECTIVES: Key tasks of survival analysis in healthcare include estimating time-to-event outcomes, comparing these across interventions, and building mathematical models for long-term projections, such as in economic evaluations. Traditionally, these tasks require switching between multiple tools, e.g., WebPlotDigitizer for digitization, R or Python for statistical modeling, and Excel for economic calculations. Data transfer, harmonization, and the need to master diverse platforms significantly increase the analysis time and create barriers for broader use, particularly among early-career researchers. The aim of this research was to develop a unified online tool enabling a complete survival analysis workflow, from data input to model export, within a single application.
METHODS: The developed tool, Surv, is implemented as a web-based application using Python. The core analytical engine is written in R, leveraging packages such as survHE, flexsurv, survival, and splines. The app also integrates a modified version of WebPlotDigitizer under an open-source license to support curve recognition from images. All components are accessible via a browser without local installation or coding.
RESULTS: Users can upload one of three data types: (1) individual patient-level time-to-event data, (2) digitized Kaplan-Meier coordinates, or (3) a Kaplan-Meier curve image. The tool supports single- or two-arm analysis. For two-arm comparisons, IPD or pseudo-IPD are used to generate Kaplan-Meier curves, hazard ratios with 95% confidence intervals, RMST, and proportional hazards testing (Schoenfeld residuals). For any number of interventions, users can perform extrapolation using standard parametric distributions (e.g., Weibull, Gompertz), piecewise models, spline-based functions, or mixture-cure models. Final models are exported to Excel with pre-loaded formulas, allowing users to select time horizon and step size for downstream economic modelling.
CONCLUSIONS: The proposed tool offers a complete, code-free survival analysis workflow in a single platform, reducing entry barriers and facilitating broader adoption among both novice and experienced researchers in outcomes and economic evaluation.