Retrospective Validation of a Stage 5 Decision Support Algorithm Using Clinical Trial Data From Stage 4 Smart Insulin Pens
Authors' conclusion
The frequency of algorithm-based coaching events served as a robust indicator of glycemic control.
Study context
- Population
- Individuals with diabetes managed with MDI in a 12-week RCT, with continuous glucose monitoring (CGM) and smart insulin pen data338 biweekly CGM profiles
- Participants
- 58 individuals
- Comparison
- Coaching event frequency in CGM profiles not meeting vs meeting CGM targets
Primary result
As reported in the abstract
Coaching event frequency by CGM target achievement · each 14-day period
- Hyperglycemia-related coaching frequency, profiles not meeting vs meeting TIR target
- 18.8 vs 34.8
- Hyperglycemia-related coaching frequency, profiles not meeting vs meeting TAR >180 mg/dL target
- 18.3 vs 33.8
- Missed bolus event frequency, TIR
- 7.6 vs 11.2
- Missed bolus event frequency, TAR
- 7.0 vs 11.1
- Higher frequencies in profiles not meeting hyperglycemia-related CGM targets
- all P < .001
- Discrimination of TIR and TAR target achievement by both coaching categories
- all AUC > 0.7
- Missed basal detection frequency associated with TIR and TAR target achievement
- all P < .05
- Hypoglycemia coaching discriminated absence of nocturnal time below range <54 mg/dL
- AUC = 0.740
- Basal insulin titration coaching frequency associated with nocturnal mean glucose changes
- all P < .005
- Real-time insulin and correction factor reduction recommendations independently associated with deterioration in TIR and TAR
- all P < .05
Original publication
Retrospective Validation of a Stage 5 Decision Support Algorithm Using Clinical Trial Data From Stage 4 Smart Insulin Pens. J Diabetes Sci Technol. Published 3 August 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →