Publication-ready dose-response and survival curves
Two analysis tools that run entirely in this browser tab: dose-response curve fitting with IC50 confidence intervals, and Kaplan-Meier survival analysis with a log-rank test. Paste a CSV, get a figure you can put in a paper.
Free, in full. No account, no upload, no subscription.
The tools
Dose-response (IC50)
Paste a concentration column and a response column. You get a fitted four-parameter logistic curve on a log-dose axis, the IC50 or EC50 with both an asymptotic and a profile-likelihood 95% confidence interval, the Hill slope, the top and bottom plateaus, R², and every fitted parameter with its standard error. Replicates become error bars; a group column becomes overlaid curves.
Open the tool →
Kaplan-Meier survival
Paste a time column, an event column and an optional group column. You get step survival curves with censoring ticks, 95% confidence bands, the number at risk under the axis, median survival per group with its confidence interval, and the log-rank chi-square, degrees of freedom and p-value for the comparison.
Open the tool →
What makes this different
Your data never leaves the browser
Everything — parsing, curve fitting, the log-rank test, the figure, the PNG — happens in JavaScript that has already been downloaded. There is no endpoint on this site that would accept a dataset. Open your network panel and watch: once the page has loaded, nothing goes out. That also means it works on a plane, and that unpublished data stays unpublished. Buying a licence is the one request this site ever makes, and it carries a plan name.
Every number has a source
The methods page gives the equation for each result, its primary reference, and the published worked example the test suite reproduces it against — including NIST certified reference values for the curve fitting, which are computed in extended precision rather than by another program. The checks run on every change.
Priced like a tool, not a tenancy
Both tools are free and complete. Exported figures carry one small credit line, and a one-time $49 licence removes it — once, not every year, on as many machines as you like. The incumbent is GraphPad Prism at $271 per seat, per year.
What this is not
It is two tools, not a statistics package. There is no ANOVA, no t-test, no Cox proportional-hazards model, no multiple-comparison correction and no data management. If you need those, you need GraphPad Prism or R, and that is a reasonable thing to need. What is here is meant to be genuinely good at two jobs that a great many people do repeatedly and reluctantly — and to be checkable, which most free alternatives are not.