| PICO | Description |
|---|---|
| Population | Adults with type 1 diabetes using multiple daily insulin injections and suboptimal glycemic control (HbA1c ≥ 7.5%). |
| Intervention | Bayesian decision support system (DSS) comprising a mobile application and insulin titration algorithm providing weekly basal and prandial insulin dose recommendations. |
| Comparison | Standard care with multiple daily insulin injections and routine clinician-managed dose adjustments without decision support system. |
| Outcome | DSS use over 12 weeks significantly improved glycemic control (reduction in HbA1c) compared to standard care. Intervention group experienced a small increase in mild hypoglycemic events but no increase in severe hypoglycemia. |
RCT
Nat Commun · 2025
Bayesian DSS for insulin dosing in type 1 diabetes
RCT · type 1 diabetes · 12 weeks
Trial design
Change from baseline — both arms
DSS HbA1c
8.6→8.1%
p=0.0002
Control HbA1c
8.6→8.5%
p=0.22
Treatment effect
-0.40%
p=0.025
Severe hypo / DKA
0
both arms
⬡ Bottom Line
A Bayesian decision support system lowered HbA1c by 0.40% versus a non-adaptive bolus calculator over 12 weeks, with no severe hypoglycemia or ketoacidosis.
P
Population
Adults with type 1 diabetes on multiple daily injections and HbA1c ≥ 7.5%.
I
Intervention
Bayesian DSS (mobile app + titration algorithm) giving weekly basal and prandial dose recommendations.
C
Comparator
Non-adaptive bolus calculator (control), both with Freestyle Libre sensors.
O
Outcome
Change in HbA1c from baseline; secondary glycemic and insulin metrics.