Online Classification of Unstructured Free-Living Exercise Sessions in People with Type 1 Diabetes
Authors' conclusion
The proposed algorithm reliably identified physical activity associated with expected change in glucose, which could be integrated into an AID system to manage the exercise disturbance in glycemia according to the predicted class.
Study context
- Population
- People with T1D; 30-min structured at-home exercise sessions (aerobic, resistance or mixed) and free-living unstructured activities
- Participants
- 1,610 structured exercise sessions
- Comparison
- Aerobic vs Interval vs Resistance
- Outcome
- Classification of exercise events (structured or unstructured) as aerobic, interval or resistance, for incorporation into an automated insulin delivery (AID) system
Methodological detailslong short-term memory network model, real-world data, triaxial accelerometer, heart rate and activity duration inputs
- Long short-term memory network model
- Real-world data
- Triaxial accelerometer, heart rate and activity duration inputs
Performance
As reported in the abstract
Classification accuracy, structured sessions, testing set
| Test or model | Value | Accuracy |
|---|---|---|
| Aerobic | 72% | 72% |
| Interval | 65% | 65% |
| Resistance | 77% | 77% |
Value
Aerobic72%
Interval65%
Resistance77%
Accuracy
Aerobic72%
Interval65%
Resistance77%
Further findings
Mean, mg/dL
| Outcome | Aerobic | Interval | Resistance |
|---|---|---|---|
| Change in glucose during unstructured sessions, by predicted class | −20.8 (SD 40.3) | −16.2 (SD 39.0) | −11.6 (SD 38.8) |
- Change in glucose during unstructured sessions, by predicted class: 3328 unstructured sessions
Original publication
Online Classification of Unstructured Free-Living Exercise Sessions in People with Type 1 Diabetes. Diabetes Technol Ther. Published 24 May 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →