Data analysis
Plan it, then analyse it
Two tools that run in your browser. Your data does not leave your computer.
Sample size and power
Calculate the sample size or the power for your study design. Seven study designs: two means, two proportions, paired measurements, non-inferiority, correlation, survival and logistic regression.
Read more
Work out how many patients your study needs, or how much power a planned sample gives you, before you collect any data. The calculator is free for everyone and runs in your browser, so no data leaves your computer.
How Datathrob Does It
- Plain-language study questionsYou choose what you want to learn, such as comparing two groups, comparing the same patients before and after, or showing a new treatment is not worse. You do not need to know the statistical test.
- Seven study designsTwo means, two proportions, paired measurements, non-inferiority, correlation, survival, and logistic regression for a risk factor.
- Guided inputsEach input explains where to find the value, such as a pilot study or published trials, with typical planning ranges.
- Sample size or powerCalculate the sample size for a target power, or the power for a fixed sample size, at your chosen significance level.
- Ready for your protocolCopy a methods sentence, save a PDF, or download a full report in Word or Markdown. The report lists the formulas and the checks run against reference software.
Data and statistical analysis
Clean your data with an audit report and run descriptive analysis free. Members add inferential, predictive, prescriptive and machine learning analyses, with no limits, downloads and a Python script export.
Read more
Upload your dataset, clean it with a full audit trail, and run the analyses your study needs. Everything runs in your browser, so your data stays on your computer. Guests can run descriptive analysis on up to 50 records and 10 variables and read the results on screen. Members have no limits and can download their results.
How Datathrob Does It
- Data cleaning with an audit reportMissing values, duplicates, outliers and implausible clinical values are flagged, and every change is recorded. Download the audit report as Word, PDF or Markdown.
- Descriptive and inferential analysisSummary tables, group comparisons, chi-square tests, correlation, paired tests and Kaplan-Meier survival.
- Predictive modelsLinear, logistic and Cox regression.
- Prescriptive and advanced analysisNumber needed to treat, decision curve analysis and cost-effectiveness, plus elastic net, principal components, clustering, random forest and gradient boosting.
- Reproducible outputMembers can export a Python script that repeats every analysis.