Dataset Builder
Turn a cohort into an analysis-ready table: choose the variables, add the scores you need and clean the data, without code.
Clinical and genomic variables
Demographics, diagnoses, prescriptions, procedures, measurements, observations and death, with date and period options, plus genotypes and polygenic risk scores.
Scores and indices
Charlson and Elixhauser comorbidity indices, BMI, eGFR, ABCD-GENE, CHA2DS2-VASc, and MELD, MELD-Na and MELD 3.0.
Point-and-click cleaning
Rename, recode, bucket and drop columns, write if-then rules and arithmetic between columns, and impute missing values with mean, median, KNN, iterative or model-based methods.
Matching and your own code
Match groups on propensity scores. If you need a step COMPASS doesn't have, add your own Python transform to the dataset.
Explore before you analyze
Each column gets a profile with types, missing values and summary statistics, data-quality alerts with suggested fixes, and violin, box, bar and scatter plots.