Dexcom G6's Urgent Low Soon alert may cut time below 54 mg/dL

Dexcom-authored analyzes (1,424 and 15,000 users) link the 20-minute predictive low alert to less hypoglycemia; an independent 24-person crossover RCT found a reduction only below 2.8 mmol/L.

ⓘ

AI system

System

Dexcom G6 real-time continuous glucose monitoring system, 'Urgent Low Soon' predictive low glucose alert (also present on the G7 in one independent study)

Model

Not assessed

Release

Exact release not established.

Application developer

Dexcom, Inc.

Role

Not assessed

Configuration

The vendor analyses and PACE trial used Dexcom G6 devices. The Preechasuk study used Dexcom G7 devices.

Inputs and outputs

Not assessed

Autonomy

Not assessed

Workflow

Real-time CGM with the ULS alert enabled (over 93% of G6 app users in Dexcom's sample). Users who respond to the alert by treating or adjusting insulin, which the sources assume rather than measure. A low-threshold alert also set (70 or 80 mg/dL in the Dexcom analysis).

Human oversight

Not assessed

Relation to AI

The alert is generated when the G6 algorithm forecasts glucose at or below 55 mg/dL within 20 minutes. Users still have to act on it. The principal outcome evidence is the maker's own real-world analyses (Puhr et al. 2019, 2020). They compare the same users before and after the G6 transition without a control arm. The independent Imperial College PACE randomized trial found no difference at 3.0 or 3.9 mmol/L after exercise and a difference only below 2.8 mmol/L. The Imperial switch study cannot separate the alert from the change of device. None of the sources describes the prediction method as machine learning.

AI contribution

The alert is generated when the G6 algorithm forecasts glucose at or below 55 mg/dL within 20 minutes. Users still have to act on it. The principal outcome evidence is the maker's own real-world analyses (Puhr et al. 2019, 2020). They compare the same users before and after the G6 transition without a control arm. The independent Imperial College PACE randomized trial found no difference at 3.0 or 3.9 mmol/L after exercise and a difference only below 2.8 mmol/L. The Imperial switch study cannot separate the alert from the change of device. None of the sources describes the prediction method as machine learning.

Benefits

Outcome identifiers

Not assessed

Benefit

time below 54 mg/dL fell from 1.0% to 0.6%

Category

Improved Physical Health

Status

Not assessed

Beneficiary groups

People. Organizations.

Beneficiaries

Not assessed

Reach

Not assessed

Distribution and access

Not assessed

Duration

Not assessed

Expected?

Yes

Costs and harms

Not assessed

Measurement

Studies

Not assessed

Study design

Not assessed

Measurement

Not assessed

Setting and sample

Not assessed

Measures

time below 54 mg/dL fell from 1.0% to 0.6%

Comparison

the same users on the preceding Dexcom G5 system, which had no predictive alert (before-and-after)

Results

Reference: 1.0 (1.5) percent of time <54 mg/dL on G5 (70 mg/dL low-alert setting, Table 1). Result: 0.6 (1.0) percent of time <54 mg/dL on G6 (70 mg/dL low-alert setting, Table 1).

Additional value from AI

Not assessed

Limitations

Not assessed

Denominators

Not assessed

Materials for replication

Not assessed

Money

Kind of value

Not assessed

Amount

Not assessed

Currency and prices

Not assessed

Period and unit

Not assessed

Realization

Not assessed

Recipients and payers

Not assessed

Gross or net

Not assessed

Included costs

Not assessed

Comparison

Not assessed

Method

Not assessed

Return and payback

Not assessed

People and places

Participants

Developer: Dexcom, Inc.

Jurisdictions

Benefit: United States (vendor analyzes); London, United Kingdom (independent trials). Developer: United States.

Ownership and control

Not assessed

Dependencies

Not assessed

Cross-border connections

Not assessed

Geopolitical implications

Not assessed

Follow-up

Conditions for replication

Users keep the alert on rather than muting it for alarm fatigue, as the Prague patient sometimes did. Sensor accuracy sufficient to avoid nuisance alerts, which Puhr et al. cite as a cause of CGM discontinuation. Benefit beyond a plain threshold alert, which the PACE trial found only below 2.8 mmol/L.

Replication attempts

3 follow-up entries. Benefit recurred: 2. Benefit did not recur: 1.

Later findings

Not assessed

Availability

Not assessed

Institutional decision

Not assessed

Grounds

Not assessed

Action and result

Not assessed

Success criterion

Not assessed

Evidence

Source numbers

AIBD-0295-source-0001–0006

Citations

Numbered references below.

Language

Not assessed

Source relationships

FDA cleared the G6 after reviewing its performance. The later studies assess hypoglycemia. The controlled PACE trial found no difference on its primary comparison and a benefit on a secondary measure. [3] [4]

Original evidence

FDA cleared the G6 after reviewing its performance. The later studies assess hypoglycemia. The controlled PACE trial found no difference on its primary comparison and a benefit on a secondary measure. [3] [4]

Access and versions

Not assessed

Supported findings

Not assessed

Evidence strength

Not assessed

Missing evidence

Not assessed

Grades

Assessed benefits

People. Organizations.

Depth

Not assessed

Reasons

People: Reductions in time below 54–55 mg/dL are reported for user populations, mostly in Dexcom's own analyses. No individual event with a changed outcome is documented. The one independent case report found ULS alone insufficient. Organizations: No payer, clinic or health-system outcome is reported in the sources.

Observed or intended

Not assessed

Rubric

Not assessed

Assessor and date

Not assessed

Alternative assessments

Not assessed

Fill

Independently studied

Entry details

Kind

Not assessed

Field

Healthcare (diabetes technology). Diabetes technology.

Purpose

Built to warn users 20 minutes before predicted serious hypoglycemia so they can act. The reported reductions in time below 54–55 mg/dL are that purpose.

Dates

Date basis unassessed: 2019 (precision unassessed). Updated: 25 October 2026.

Stage

Not assessed

Version

SHA-256: a22b940e01e8…

References

Other identifiers

Not assessed

Category mappings

Not assessed

Brief

Not assessed

Revisions

8 revisions. Latest: 25 October 2026.

Sources

  1. Peer-reviewed study Developer

    Puhr S, Derdzinski M, Welsh JB, Parker AS, Walker T, Price DA (2019). Real-World Hypoglycemia Avoidance with a Continuous Glucose Monitoring System's Predictive Low Glucose Alert. Diabetes Technology & Therapeutics 21(4):155-158 (PMC6477579). All authors at Dexcom, Inc.

  2. Peer-reviewed study Developer

    Puhr S, Derdzinski M, Parker AS, Welsh JB, Price DA (2020). Real-World Hypoglycemia Avoidance With a Predictive Low Glucose Alert Does Not Depend on Frequent Screen Views. Journal of Diabetes Science and Technology 14(1) (online 3 April 2019; PMC7189147). All authors at Dexcom, Inc.

  3. Regulatory document Independent author Context or contrary evidence

    U.S. Food and Drug Administration (2018). Evaluation of Automatic Class III Designation for Dexcom G6 Continuous Glucose Monitoring System, Decision Summary, DEN170088 (decision date 27 March 2018).

    Sensor-performance evidence, the alert’s intended function, and the clearance decision. § L(3), pp. 13–14; §§ S–T, pp. 38–39.

    FDA’s regulatory review predates the outcome studies.

  4. Peer-reviewed study Independent author

    Rilstone S, Oliver N, Godsland I, Tanushi B, Thomas M, Hill N (2024). A Randomized Controlled Trial Assessing the Impact of Continuous Glucose Monitoring with a Predictive Hypoglycemia Alert Function on Hypoglycemia in Physical Activity for People with Type 1 Diabetes (PACE). Diabetes Technology & Therapeutics 26(2); doi:10.1089/dia.2023.0376.

    Original randomized study; distinct from clearance. A controlled comparison of alerts on and off; the primary hypoglycemia comparison was null, with a favorable secondary result. Study design, Results, and Discussion.

    Original randomized study; distinct from clearance.

  5. Peer-reviewed study Independent author

    Preechasuk L, Avari P, Oliver N, Reddy M (2024). Switching from Intermittently Scanned Continuous Glucose Monitoring to Real-Time Continuous Glucose Monitoring with a Predictive Urgent Low Soon Alert Reduces Exposure to Hypoglycemia. Diabetes Technology & Therapeutics 26(7); doi:10.1089/dia.2023.0434.

Collections: Scrutiny and findings, Cross-border cases.