Classifying Continuous Glucose Monitoring Documents From Electronic Health Records
Authors' conclusion
Nearly half of CGM-related documents were AGP reports, which are useful for clinical practice and diabetes research; however, the remaining half are other clinical documents.
Study context
- Population
- Continuous glucose monitoring (CGM)-related documents in electronic health records (EHR), NYU Langone Health
- Participants
- 2,244 documents
- Comparison
- Document classification algorithm vs expert manual review
- Manual reviewn = 62reference
- Outcome
- Sensitivity (correct classification of CGM AGP report) and specificity (correct classification of non-CGM report) of the algorithm vs manual review
Methodological detailsrandom sample of EHR documents, independent manual review by two experts
- Random sample of EHR documents
- Independent manual review by two experts
Performance
As reported in the abstract
Algorithm vs manual review
| Test or model | Sensitivity | Specificity |
|---|---|---|
| Classification algorithm | 95.0% | 91.7% |
Classification algorithm
Sensitivity95.0%
Specificity91.7%
Further findings
Document classification
- CGM Ambulatory Glucose Profile (AGP) reports (43.3% FreeStyle Libre, 56.7% Dexcom)
- 1040 (46.5%)
- Non-CGM reports
- 1170 (52.1%)
- Uncertain documents
- 34 (1.5%)
Inter-expert agreement
- Agreement for sensitivity
- 100%
- Agreement for specificity
- 98.4%
Original publication
Classifying Continuous Glucose Monitoring Documents From Electronic Health Records. J Diabetes Sci Technol. Published 12 March 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →