Long-Term Prediction of Blood Glucose Levels in Type 1 Diabetes Using a CNN-LSTM-Based Deep Neural Network
Authors' conclusion
Our investigation demonstrated that our method achieved superior glucose forecasting compared with existing approaches in the literature, and thanks to its generalizability showed potential for real-life applications.
Study context
- Population
- People with T1D; Blood glucose (BG) prediction from historical glucose measurements, meal information and insulin intakes
- Comparison
- Replace-BG vs DIAdvisor
- Outcome
- BG prediction accuracy, 90-minute prediction horizon
- Follow-up
- 90-minute PH
Methodological detailsdeep learning, CNN-LSTM
- Deep learning
- CNN-LSTM
Performance
As reported in the abstract
| Test or model | Mean absolute error | Root mean square error | Coefficient of determination | CG-EGA accurate predictions | CG-EGA benign errors | CG-EGA erroneous predictions |
|---|---|---|---|---|---|---|
| Replace-BG | 17.30 ± 2.07 mg/dL | 23.45 ± 3.18 mg/dL | 84.13 ± 4.22% | 94.71 ± 3.89% | 1.81 ± 1.06% | 3.47 ± 1.12% |
| DIAdvisor | 18.23 ± 2.97 mg/dL | 25.12 ± 4.65 mg/dL | 82.34 ± 4.54% | 91.71 ± 4.32% | 2.51 ± 0.86% | 5.78 ± 1.72% |
Mean absolute error
Replace-BG17.30 ± 2.07 mg/dL
DIAdvisor18.23 ± 2.97 mg/dL
Root mean square error
Replace-BG23.45 ± 3.18 mg/dL
DIAdvisor25.12 ± 4.65 mg/dL
Coefficient of determination
Replace-BG84.13 ± 4.22%
DIAdvisor82.34 ± 4.54%
CG-EGA accurate predictions
Replace-BG94.71 ± 3.89%
DIAdvisor91.71 ± 4.32%
CG-EGA benign errors
Replace-BG1.81 ± 1.06%
DIAdvisor2.51 ± 0.86%
CG-EGA erroneous predictions
Replace-BG3.47 ± 1.12%
DIAdvisor5.78 ± 1.72%
Original publication
Long-Term Prediction of Blood Glucose Levels in Type 1 Diabetes Using a CNN-LSTM-Based Deep Neural Network. J Diabetes Sci Technol. Published 25 April 2022. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →