Kaplan-Meier Curve Maker

The required input is one row per participant with a non-negative follow-up time and an event status. Add an optional group column for treatment or cohort comparisons. Before analysis, confirm whether 1 means event and 0 means censored, or whether your dataset uses the reverse convention.

Either convention can appear in real datasets, so you must confirm it explicitly. Select 1 = event and 0 = censored, or reverse the mapping when that matches your source data. The parser also recognizes words such as event, dead, alive, and censored, but you should still review the result.

Copy the header row and participant rows from Excel or Google Sheets, then paste them into the data field. Keep time and event as separate columns and add group when needed. You can also upload CSV or TSV. The browser calculates and redraws the curve without requiring R, Python, or desktop software.

A censoring mark appears on the curve at the survival probability current when that participant’s follow-up ended. Censoring removes the participant from later risk sets but does not create a downward survival step. When events and censoring share a time, events are handled before censored records leave the risk set.

Median survival is the first time the estimated survival probability reaches or falls below 0.50. If the curve remains above 0.50 throughout observed follow-up, the median is not reached. Report it as not reached rather than substituting the ordinary median follow-up time or extending the curve beyond the observed data.

At each displayed time point, the table counts participants whose recorded follow-up time is at least that time. This is the risk set immediately before events at the selected point. The table uses the complete dataset, so shortening the visible x-axis does not delete later follow-up or alter earlier risk counts.

Add a group column containing two or more labels. The generator draws one curve per group and reports a global log-rank test with k−1 degrees of freedom when the covariance matrix is estimable. It does not report pairwise tests or adjusted hazard ratios, which require a separate prespecified analysis.

The tool parses, analyzes, and draws participant rows in your browser rather than uploading them. Export SVG or high-resolution PNG for figure preparation and keep the statistics CSV for review. Remove direct identifiers, follow institutional data policy, and independently verify the analysis, labels, assumptions, and target-journal requirements before publication.