Trial Outcomes & Findings for The Pediatric Artificial Pancreas Automated Initialization Trial (NCT NCT06017089)

NCT ID: NCT06017089

Last Updated: 2025-09-15

Results Overview

% of time below 54 mg/dL

Recruitment status

COMPLETED

Study phase

NA

Target enrollment

33 participants

Primary outcome timeframe

Baseline and Weeks 1-8

Results posted on

2025-09-15

Participant Flow

Did not meet screening eligibility (N=1)

Participant milestones

Participant milestones
Measure
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
In this single-arm intervention trial, all participants will use the study system (t:slim X2 with Control-IQ Technology and Dexcom Continuous Glucose Monitor) in closed-loop mode for 8 weeks at home with periodic parameter adjustment driven by an AI-based Advisor system. AI-based Advisor system: Tandem t:slim X2 with Control-IQ and t:connect mobile application and Dexcom G6 or G7 system, connected to University of Virginia (UVA) cloud-based Physician Dashboard with insulin pump parameters driven by an AI-based Advisor system.
Overall Study
STARTED
32
Overall Study
COMPLETED
28
Overall Study
NOT COMPLETED
4

Reasons for withdrawal

Withdrawal data not reported

Baseline Characteristics

The Pediatric Artificial Pancreas Automated Initialization Trial

Baseline characteristics by cohort

Baseline characteristics by cohort
Measure
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=32 Participants
In this single-arm intervention trial, all participants will use the study system (t:slim X2 with Control-IQ Technology and Dexcom Continuous Glucose Monitor) in closed-loop mode for 8 weeks at home with periodic parameter adjustment driven by an AI-based Advisor system. AI-based Advisor system: Tandem t:slim X2 with Control-IQ and t:connect mobile application and Dexcom G6 or G7 system, connected to UVA cloud-based Physician Dashboard with insulin pump parameters driven by an AI-based Advisor system.
Age, Customized
2 to <4 years
9 Participants
n=5 Participants
Age, Customized
4 to <6 years
23 Participants
n=5 Participants
Sex: Female, Male
Female
11 Participants
n=5 Participants
Sex: Female, Male
Male
21 Participants
n=5 Participants
Ethnicity (NIH/OMB)
Hispanic or Latino
5 Participants
n=5 Participants
Ethnicity (NIH/OMB)
Not Hispanic or Latino
27 Participants
n=5 Participants
Ethnicity (NIH/OMB)
Unknown or Not Reported
0 Participants
n=5 Participants
Race (NIH/OMB)
American Indian or Alaska Native
0 Participants
n=5 Participants
Race (NIH/OMB)
Asian
3 Participants
n=5 Participants
Race (NIH/OMB)
Native Hawaiian or Other Pacific Islander
0 Participants
n=5 Participants
Race (NIH/OMB)
Black or African American
3 Participants
n=5 Participants
Race (NIH/OMB)
White
22 Participants
n=5 Participants
Race (NIH/OMB)
More than one race
4 Participants
n=5 Participants
Race (NIH/OMB)
Unknown or Not Reported
0 Participants
n=5 Participants
Body Mass Index Percentile
76 percentage
n=5 Participants
Parent Education
High school graduate/diploma/GED
4 Participants
n=5 Participants
Parent Education
Technical/Vocational
1 Participants
n=5 Participants
Parent Education
Associate Degree (AA)
4 Participants
n=5 Participants
Parent Education
College Graduate (Bachelor's Degree or Equivalent)
9 Participants
n=5 Participants
Parent Education
Advanced Degree (e.g. Masters, PhD, MD)
14 Participants
n=5 Participants
Annual Household Income
$35,000 to <$50,000
1 Participants
n=5 Participants
Annual Household Income
$50,000 to <$75,000
5 Participants
n=5 Participants
Annual Household Income
$75,000 to <$100,000
4 Participants
n=5 Participants
Annual Household Income
$100,000 to <$200,000
10 Participants
n=5 Participants
Annual Household Income
≥$200,000
11 Participants
n=5 Participants
Annual Household Income
Unknown
1 Participants
n=5 Participants
Health Insurance
Private
26 Participants
n=5 Participants
Health Insurance
Medicare
2 Participants
n=5 Participants
Health Insurance
Medicaid
4 Participants
n=5 Participants
HbA1c
7.3 percentage of glycated hemoglobin
STANDARD_DEVIATION 0.9 • n=5 Participants
Insulin Modality
MDI
20 Participants
n=5 Participants
Insulin Modality
Pump
12 Participants
n=5 Participants
Total Daily Insulin
0.61 U/kg/day
STANDARD_DEVIATION 0.21 • n=5 Participants
Number of Pump Boluses or Injections of Short-Acting Insulin per Day
5 Pump boluses per day
n=5 Participants
Diabetic Ketoacidosis (DKA) in last 12 months
No episodes
24 Participants
n=5 Participants
Diabetic Ketoacidosis (DKA) in last 12 months
One episodes
8 Participants
n=5 Participants
Severe hypoglycemia (SH) in Last 12 Months
None
32 Participants
n=5 Participants
Severe hypoglycemia (SH) in Last 12 Months
1
0 Participants
n=5 Participants
Diabetes Duration
10 months
n=5 Participants

PRIMARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

% of time below 54 mg/dL

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
Safety Endpoint (Tested for Non-inferiority Compared to Baseline) CGM Measured (a)
0.3 percentage of time
Standard Deviation 0.3
0.3 percentage of time
Standard Deviation 0.2

PRIMARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

% of time above 250 mg/dL

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
Safety Endpoint (Tested for Non-inferiority Compared to Baseline) CGM Measured (b)
13 percentage of time
Standard Deviation 10
9 percentage of time
Standard Deviation 5

PRIMARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

% of time in range 70-180 mg/dL

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
Hierarchical Efficacy Endpoints (Tested for Superiority Compared With Baseline) CGM Measured (a)
63 percentage of time
Standard Deviation 15
70 percentage of time
Standard Deviation 8

PRIMARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

Mean glucose

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
Hierarchical Efficacy Endpoints (Tested for Superiority Compared With Baseline) CGM Measured (b)
168 mg/dl
Standard Deviation 28
157 mg/dl
Standard Deviation 14

PRIMARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

% of time \>250 mg/dL

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
Hierarchical Efficacy Endpoints (Tested for Superiority Compared With Baseline) CGM Measured (c)
13 percentage of time
Standard Deviation 10
9 percentage of time
Standard Deviation 5

PRIMARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

% of time \<70 mg/dL

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
Hierarchical Efficacy Endpoints (Tested for Superiority Compared With Baseline) CGM Measured (d)
2 percentage of time
Standard Deviation 1.4
1.7 percentage of time
Standard Deviation 0.9

PRIMARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

% of time \<54 mg/dL

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
Hierarchical Efficacy Endpoints (Tested for Superiority Compared With Baseline) CGM Measured (e)
0.3 percentage of time
Standard Deviation 0.3
0.3 percentage of time
Standard Deviation 0.2

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

% of time spent within range 70 mg/dL-140 mg/dL.

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
CGM Measured Time in Range
42 percentage of time
Standard Deviation 14
47 percentage of time
Standard Deviation 9

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

% of time \>180 mg/dL

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
CGM Measured (a)
35 percentage of time
Standard Deviation 15
29 percentage of time
Standard Deviation 8

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

% of time \>300 mg/dL

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
CGM Measured (b)
5.9 percentage of time
Standard Deviation 5.6
3.3 percentage of time
Standard Deviation 2.6

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

% of time \<60 mg/dL

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
CGM Measured (c)
0.7 percentage of time
Standard Deviation 0.5
0.6 percentage of time
Standard Deviation 0.4

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

Glucose standard deviation

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
CGM Measured (d)
64 mg/dl
Standard Deviation 16
59 mg/dl
Standard Deviation 12

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

Glucose coefficient of variation

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
CGM Measured (e)
38 percent
Standard Deviation 5
37 percent
Standard Deviation 5

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

The high blood glucose index (HBGI) is based on a nonlinear transformation of blood glucose values that corrects for the asymmetry of the glucose scale. This transformation maps glucose values into a risk space (minimum risk = 0), where higher values correspond to higher risk. Values below 10 suggest low to moderate risk.

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
CGM Measured (f)
8.8 index
Standard Deviation 5
6.8 index
Standard Deviation 2.5

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

The low blood glucose index (LBGI) is based on a nonlinear transformation of blood glucose values that corrects for the asymmetry of the glucose scale. This transformation maps glucose values into a risk space (minimum risk = 0), where higher values correspond to higher risk. Values \<1 suggest low risk of hypoglycemia.

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
CGM Measured (g)
0.6 index
Standard Deviation 0.4
0.6 index
Standard Deviation 0.2

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

Weekly hyperglycemic event rate

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
CGM Measured (h)
1.9 events
Standard Deviation 2.0
1.1 events
Standard Deviation 1.0

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

Weekly hypoglycemic event rate

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
CGM Measured (i)
0.6 events
Standard Deviation 0.6
0.4 events
Standard Deviation 0.4

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

Number of participants whose % of time in range 70-180 mg/dL improved by 5% or more from baseline to 8 weeks.

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
Binary Outcome 1
13 Participants

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

Number of participants whose % of time in range 70-180 mg/dL improved by 10% or more from baseline to 8 weeks.

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
Binary Outcome 2
9 Participants

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

Participants with % of time in range 70-180 mg/dL \>70% and % of time \<70 mg/dL \<4%

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=29 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
Binary Outcome 3
10 Participants
13 Participants

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

Total daily insulin (units/kg)

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=27 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
Total Daily Insulin
0.59 U/kg
Standard Deviation 0.20
0.66 U/kg
Standard Deviation 0.19

SECONDARY outcome

Timeframe: Baseline and Weeks 1-8

Population: Three participants were excluded due to lack of data. To meet criteria for inclusion, participants had to provide at least 168 hours of CGM data during baseline and the 8-week follow-up period and remain in closed loop for at least 50% of the 8-week period after closed loop initiation.

Percentage of total insulin delivered via basal administration.

Outcome measures

Outcome measures
Measure
Baseline Period
n=29 Participants
Participants aged 2-6 years with Type 1 Diabetes were monitored using their existing insulin therapy (either multiple daily injections or personal insulin pump settings) prior to the initiation of the AI-driven intervention. Continuous glucose monitoring (CGM) data were collected for up to 28 days before the intervention to establish baseline glycemic control metrics.
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=27 Participants
Participants used the Tandem t:slim X2 insulin pump with Control-IQ technology for 8 weeks. Initial and adaptive pump settings were guided by an AI-based advisor through the UVA Clinical Portal. Investigators reviewed and could override AI-generated recommendations. This period was designed to evaluate the safety and efficacy of algorithm-driven insulin delivery in a pediatric population.
Basal Insulin
39 percentage of total insulin
Standard Deviation 17
38 percentage of total insulin
Standard Deviation 9

Adverse Events

AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation

Serious events: 3 serious events
Other events: 5 other events
Deaths: 0 deaths

Serious adverse events

Serious adverse events
Measure
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=32 participants at risk
In this single-arm intervention trial, all participants will use the study system (t:slim X2 with Control-IQ Technology and Dexcom Continuous Glucose Monitor) in closed-loop mode for 8 weeks at home with periodic parameter adjustment driven by an AI-based Advisor system. AI-based Advisor system: Tandem t:slim X2 with Control-IQ and t:connect mobile application and Dexcom G6 or G7 system, connected to UVA cloud-based Physician Dashboard with insulin pump parameters driven by an AI-based Advisor system.
Endocrine disorders
Severe Hypoglycemia
3.1%
1/32 • Number of events 1 • 8 weeks
Endocrine disorders
Diabetic Ketoacidosis
3.1%
1/32 • Number of events 1 • 8 weeks
Endocrine disorders
Other serious adverse events
3.1%
1/32 • Number of events 1 • 8 weeks

Other adverse events

Other adverse events
Measure
AI Advisor-driven At-home Closed Loop System Initiation and Parameter Adaptation
n=32 participants at risk
In this single-arm intervention trial, all participants will use the study system (t:slim X2 with Control-IQ Technology and Dexcom Continuous Glucose Monitor) in closed-loop mode for 8 weeks at home with periodic parameter adjustment driven by an AI-based Advisor system. AI-based Advisor system: Tandem t:slim X2 with Control-IQ and t:connect mobile application and Dexcom G6 or G7 system, connected to UVA cloud-based Physician Dashboard with insulin pump parameters driven by an AI-based Advisor system.
Endocrine disorders
Hyperglycemia with or without Ketosis
9.4%
3/32 • Number of events 3 • 8 weeks
Gastrointestinal disorders
Gastroenteritis
6.2%
2/32 • Number of events 2 • 8 weeks

Additional Information

Marc Breton, PhD

University of Virginia Center for Diabetes Technology

Phone: 434-982-6484

Results disclosure agreements

  • Principal investigator is a sponsor employee
  • Publication restrictions are in place