Missing or inconsistent units
Standardize unit-of-measure fields so results can be evaluated within a consistent structure.
Case study · ALT/AST data harmonization
In an ALT/AST source sample, Prognos harmonization raised numeric-field completeness from 2.1% to 99.9%, making more records usable for analysis.
One result. More usable context.
0.0000^44.0000H^Field-completeness results from the evaluated ALT/AST source sample.
The challenge
The sample’s result-value field was already 99.9% complete. Yet only 2.1% of records had all numeric fields populated.
Missing units, packed reference ranges, and inconsistent abnormal flags left important context difficult to use consistently. A high count of test results alone did not establish whether the records were ready for analysis.
The opportunity: make the information surrounding each result usable through validated source interpretation and consistent field structure.
What Prognos addressed
Standardize unit-of-measure fields so results can be evaluated within a consistent structure.
Parse source range strings into separate lower and upper bounds.
Normalize source patterns into consistent abnormal-result indicators.
The results
Harmonization increased the share of records with all numeric fields populated by 97.8 percentage points in the evaluated sample.
The largest improvements were in units and abnormal flags. Result values, reference ranges, and LOINC completeness remained at their reported baseline levels.
Share of records in the evaluated ALT/AST sample
| Evaluated field | Raw lab data | Prognos data |
|---|---|---|
| Abnormal flag | 10.2% | 99.9% |
| Unit of measure | 21.9% | 100% |
| Result value | 99.9% | 99.9% |
| Reference range | 100% | 100% |
| All numeric fields populated | 2.1% | 99.9% |
| LOINC | 100% | 100% |
The reported HCP completeness measure remained 96% before and after processing.
The source separately reports an increase in secondary-specialty population, adding provider context within this sample.
The approach
Prognos applied source-specific interpretation to standardize the fields around ALT and AST results.
Populate consistent unit fields using validated source interpretation. In the examples, source units appear as null or IU/L; standardized units appear as U/L.
Separate packed reference-range strings into numeric lower and upper bounds that can be used in analysis.
Convert source patterns such as H^ into H, and populate normal indicators where supported by the validated interpretation.
Inside the records
Explore the four record examples from the source analysis. Each pairs the original fields with the harmonized output.
Raw laboratory record
Harmonized Prognos record
In this sample, the value came from improving the completeness and structure of result context around existing lab values.
Explore Prognos
Discuss the tests, source formats, and data requirements that matter for your program.
Talk with Prognos about the completeness, consistency, and clinical context your analysis requires.