Efficacy of an AI-Enabled Low Glucose Prediction: A Pooled Performance Analysis With Capillary Blood Glucose as Ground Truth
Authors' conclusion
The LGP feature is an accurate, highly sensitive, and specific tool for timely, proactive low glucose prediction, validated against both capillary BG and CGM. This predictive intelligence is a crucial mechanism for people with diabetes to safely mitigate hypoglycemia risk, addressing a significant clinical gap and potentially reducing fear of hypoglycemia and diabetes distress.
Study context
- Population
- Insulin-dependent people with diabetes using multiple daily injections (MDI) and continuous glucose monitoring (CGM)2,709 recording days
- Participants
- 85 people
- Comparison
- Low Glucose Predict (LGP) prediction vs capillary blood glucose (BG) and CGM reference
- Outcome
- Prediction performance (sensitivity, specificity, ROC-AUC) against capillary BG and CGM
Methodological detailsretrospective pooled analysis of three prospective trials
- Retrospective pooled analysis of three prospective trials
Performance
As reported in the abstract
Performance vs capillary BG reference
| Test or model | Sensitivity | Specificity | ROC-AUC |
|---|---|---|---|
| LGP | 87.13% | 97.43% | 0.9787 |
LGP
Sensitivity87.13%
Specificity97.43%
ROC-AUC0.9787
Further findings
Performance vs capillary BG including close-call events (+10 mg/dL above threshold)
- Sensitivity
- 91.89%
- Specificity
- 98.09%
Performance vs CGM reference
- Sensitivity
- 94.40%
- Specificity
- 98.25%
Mean lead time (CGM reference)
- Mean lead time
- 14.71 ± 8.30 minutes
Average daily true notification rate
- Average daily true notifications
- 1.31
- Including close-calls
- 2.60
Original publication
Efficacy of an AI-Enabled Low Glucose Prediction: A Pooled Performance Analysis With Capillary Blood Glucose as Ground Truth. J Diabetes Sci Technol. Published 18 January 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →