Statistical software for biomedical research, including ROC curve analysis.

MedCalc is a statistical software package for the biomedical sciences. It has an integrated spreadsheet which can be configured to contain up to 676 columns and 100000 rows. The program correctly handles missing data and provides reliable date arithmetic. MedCalc can import Excel, SPSS, DBase and Lotus files, and files in SYLK, DIF or text format. Comprehensive on line help is available and, in addition to a manual in PDF format, a complete user manual is available on the web. Special features of the program are, amongst others, lots of graphs including Kaplan-Meier survival plots, Bland & Altman plot, Deming and Passing & Bablok regression for method comparison. MedCalc also includes a complete module for Receiver Operating Characteristic (ROC) curve analysis, with calculation of sensitivity, specificity, likelihood ratios, and positive and negative predictive values for all possible threshold values. The program calculates the area under the curve (AUC) with its standard error and 95% confidence interval. Furthermore, threshold values can be selected in an interactive dot diagram with automatic calculation of corresponding sensitivity and specificity. Comparison of ROC curves includes calculation of the difference between the areas under the curves, with standard error, 95% confidence interval and P-value.