Data Quality
Quality You Can Measure, Trust You Can Prove.

Five Levels of
Data Quality Assurance

Conformance
Automated checks verifying that data aligns with the expected structure, data types, and OMOP CDM rules — ensuring it has been captured, formatted, and stored correctly.

Completeness
Verifying that information expected for a patient, encounter, or variable is actually present and sufficiently populated in the dataset.

Plausibility
Assessing whether recorded values are logical and clinically credible — either against expected ranges from the literature, or through internal consistency across variables.

Medical Review
Validation by clinical experts who assess edge cases, ambiguous records, and NLP-extracted variables, flagged by Sentinel, with real-world scrutiny.

Benchmark
Comparing dataset characteristics against known epidemiological benchmarks to detect systematic biases or site-specific anomalies.
Scalable Data Validation
With Sentinel
1 HOUR
Per cohort
Systematic and automated
Manually
6 DAYS
Per cohort
Manual Review
Sentinel’s automated pipeline removes manual validation bottlenecks, running systematic checks across thousands of data points per hospital. It detects missing values, implausible measurements, and mis-extracted clinical data, letting human experts focus only on flagged issues. The result: full, reproducible validation in just an hour per cohort — enabling data quality checks at scale.





