Skip to content
type1.science

Accu-Chek SmartGuide

Roche Diabetes Care

What it is

Roche’s adult CGM combines up to 14-day sensor wear with glucose-prediction features. It requires early fingerstick calibration. The original accuracy study reported 9.2% MARD against a capillary meter; prediction-model validation and simulated benefits are separate evidence. Current UK pages describe mySugr integration.

Editorial review: .

Most recent recorded citation date: 2026-09-15. Only explicit date metadata is included; an undated citation may be newer. This does not mean every claim was reviewed on that date.

Trial status, labels and access can change between reviews. How we review the evidence · How to read the evidence

Full sources ↓Full discussion ↓Report an issue →

Available nowRegulator-approvedcgmpredictiveaiOfficial site ↗

Evidence behind this assessment

Key evidence notes. Study results, product eligibility and access answer different questions.

Who was studied?
Study populations and analysis groups vary. Product age limits alone do not describe who was studied.See the linked studies and their populations →
Benefit or performance
Accuracy: The original study reported 9.2% MARD versus capillary blood-glucose measurements. Retrospective processing and differing study methods prevent a direct ranking against another product’s published MARD.[2]
Important harms and treatment burden
Read the safety discussion and original sources. A missing summary does not establish safety.
Approval and country access
Country-specific approval and access are not summarized in this record.Approval, trial recruitment, local supply and funding are separate. Check the cited label or access source.
Follow-up and remaining uncertainty
Read the full discussion and original sources for follow-up duration and study limitations.

This record cites company or conference reports. Those sources may describe interim findings; their source type is shown in the source list.

Editorial score: calculation and evidence

A weighted editorial judgment on a 0–100 scale, not a probability of success or a measured treatment effect. Higher criterion scores mean more favorable assessments.

Score limitations. Scores are editorial judgments, not validated comparative measurements. The sensor study used a capillary meter and retrospectively processed signals; prediction accuracy is not proof of clinical benefit. No measured device-specific lag supports the previous lag rating, which is omitted. Missing lag evidence is unknown, not zero, and reduces scoring coverage.

Default calculation: 80 × 20 + 25 × 15 + 80 × 10 + 65 × 10 + 90 × 10 + 70 × 10 + 0 × 5 + 45 × 5 = 5250; divide by total weight 85. Unrounded weighted result: 61.76470588235294.

Accuracy80

The original study reported 9.2% MARD versus capillary blood-glucose measurements. Retrospective processing and differing study methods prevent a direct ranking against another product’s published MARD.[2]

Interoperability25

Trio 1.0 lists direct SmartGuide support. This is a separate community-developed DIY integration; its release is not manufacturer or regulatory approval of an AID pairing.[12]

Sensor lifespan80

The manufacturer specifies wear up to 14 days. This is a maximum intended duration, not a promise that every sensor lasts the full period.[1]

No calibration65

First calibration is prompted after 12 hours; the second follows 30 minutes to 3 hours later. Missing the second returns the sensor to Trend Mode, whose values must not guide insulin dosing. The score reflects this additional burden.[1]

Alerts & prediction90

Retrospective validation supports prediction functionality, not prevention of every low. Pooled 30-minute low prediction had 95.2% sensitivity and 16.2-minute mean lead time; nighttime sensitivity was 55.3%. The score is an editorial feature judgment.[3]

Form factor70

The device manual specifies a sensor about 33.3 mm in diameter without adhesive and 5.9 mm high with adhesive, worn on the back of the upper arm. It is rated for temporary immersion to 1 metre for 60 minutes.[1]

Ketone sensing0

The sensor measures interstitial glucose; the manual does not specify a ketone measurement function.[1]

Access & cost45

Germany advertises prescription and self-pay ordering. This establishes a country-specific pathway, not EU-wide funding, immediate stock or eligibility for every person. Other verified country pathways are described below.[11]

The full picture

Sensor and app scope. Accu-Chek SmartGuide is an adult CGM, labeled for people with diabetes aged 18 and older. The device manual describes upper-arm wear for up to 14 days and readings every five minutes after a one-hour warm-up. Roche announced CE marking in July 2024. Current UK pages describe integrated monitoring and prediction in mySugr; the earlier SmartGuide and separate Predict apps remain the software evaluated in the original papers. Local app and phone compatibility should be checked rather than assuming one worldwide configuration.169

Calibration and backup readings. The first calibration prompt occurs 12 hours after insertion, with a second fingerstick calibration 30 minutes to 3 hours later. Missing that second step returns the sensor to Trend Mode. The instructions prohibit using Trend Mode values for insulin dosing and direct users to a meter when sensor values do not match symptoms or expectations. Predictions do not remove those requirements.1

What the accuracy study measured. The study analysed data from 48 participants and 139 sensors. Against capillary blood-glucose measurements, the clinic analysis gave 9.2% MARD with a 20/20 agreement rate of 90.5%. Reported 14-day survival was 82.4%. These abstract-level results do not support a direct accuracy ranking against another CGM’s differently designed study.2

Low readings, wear and harms. The readable abstract reports 48 participants, 139 sensors and 82.4% 14-day survival; low-range accuracy, insertion-accounting, site-event and exclusion details from the full paper were not verified here.2

Prediction is a different outcome. The model paper reported data from 21 people with T1D on injections, 226 with T1D on pumps and 59 with T2D on injections. Its pooled 30-minute low-prediction sensitivity was 95.2%, with a mean 16.2-minute warning; pooled nighttime sensitivity was 55.3% and specificity 91.6%. These are retrospective prediction metrics, not sensor MARD, guaranteed warning time or demonstrated reductions in clinical hypoglycemia. Models require internet access.3

Clinical benefit remains a separate question. A 2026 study simulated prediction responses in 204 digital twins derived from historical T1D data. Its modeled benefits are not outcomes from 204 people prospectively using the app. Two current registries have no posted results: a randomized SmartGuide-versus-SMBG study with 270 estimated participants, and a nonrandomized Predict pre/post study with 89 estimated participants. Both include T1D and insulin-treated T2D; their planned endpoints do not yet establish benefit.51314

Practical limits. The sensor’s water rating is temporary immersion to 1 metre for 60 minutes. It measures glucose, not ketones. The manual warns that oral vitamin C above 500 mg/day, any intravenous vitamin C, gentisic-acid supplements and methyldopa may falsely raise readings. It excludes critically ill and dialysis patients and requires removal before specified high-energy imaging or treatment procedures. These are selected label limits, not a substitute for the complete instructions.1

Connectivity and country access. Trio 1.0 explicitly supports SmartGuide as a community DIY integration; that release does not establish a labeled commercial AID pairing. Germany advertises prescription and self-pay ordering, while the UK/Ireland page says prescription eligibility depends on location, diabetes and treatment. Roche’s November 2025 regional announcement names Kuwait, Qatar and Saudi Arabia, with UAE then planned. These sources establish particular advertised pathways, not guaranteed stock, universal reimbursement or a complete current global country count.121198

What's next for this

  • →Randomized SmartGuide-versus-SMBG study; registry has no posted results · primary completion estimated 15 November 2026
  • →Predict pre/post behavior study; registry has no posted results · completion estimated 31 October 2026

Sources

  1. [1]
    Accu-Chek SmartGuide device user manual, English September 2025 · Manufacturer — Sections 2, 5, 6 and 10: adult indication, calibration, water limits, interfering substances and dimensions.

    Roche. SmartGuide device user manual, English September 2025, sections 2, 5, 6 and 10.

  2. [2]
    Performance of a Novel Continuous Glucose Monitoring Device in People With Diabetes · Peer-reviewed study · 2024-08-19 — Original full paper: 48 enrolled, 46 completed; capillary meter reference and retrospective signal processing. Clinic and combined datasets have different denominators.

    Mader and colleagues. Original full accuracy paper (2024), Methods, Tables 1–3, sensor survival and adverse events.

  3. [3]
    Enhancing the Capabilities of Continuous Glucose Monitoring With a Predictive App · Peer-reviewed study · 2024-08-19 — Original full paper: retrospective model validation, with separate prediction cohorts and sensitivity limits; not a prospective clinical-benefit trial.

    Herrero and colleagues. Original full prediction-model paper (2024), Methods, Tables 3 and 5, Discussion and funding.

  4. [4]
    Accu-Chek SmartGuide Predict app user manual, version 3.7.1 · Manufacturer — Product purpose, data and internet requirements; predictions do not replace the primary glucose display or control another device.
  5. [5]
    Glucose Predictions Improve Glycemic Control: A Digital Twin Evaluation · Peer-reviewed study · 2026-03-12 — Original full paper: in-silico evaluation using 204 digital twins, not 204 people prospectively treated with Predict.

    Herrero and colleagues. Original full digital-twin evaluation (12 March 2026), Methods and limitations.

  6. [6]
    Roche announces CE marking for Accu-Chek SmartGuide · Manufacturer · 2024-07-09

    Roche. CE-mark announcement (9 July 2024).

  7. [7]
    Roche announces CE-marked SmartGuide integration with mySugr · Manufacturer · 2025-09-16 — Masthead and release text say 16 September 2025. The 13-country count and expansion beyond 30 were a dated count and plan, not a verified current census.
  8. [8]
    Roche launches SmartGuide in the Middle East · Manufacturer · 2025-11-13 — The release body is dated 13 November 2025 despite the URL ending 2025-10-28. It names Kuwait, Qatar and Saudi Arabia, with UAE then planned.

    Roche. Middle East release, body dated 13 November 2025.

  9. [9]
    Current UK and Ireland SmartGuide/mySugr product and prescription information · Manufacturer

    Roche. Current UK/Ireland SmartGuide/mySugr product and prescription information, checked 17 September 2026.

  10. [10]
  11. [11]
    Germany: SmartGuide prescription and self-pay ordering · Manufacturer

    Roche. German prescription and self-pay ordering pathway, checked 17 September 2026.

  12. [12]
    Trio v1.0 official release notes · Open-source community · 2026-09-15

    Trio maintainers. v1.0 release (15 September 2026).

  13. [13]
    SmartGuide versus self-monitoring of blood glucose (NCT06704672) · Trial registry · 2026-08-28

    ClinicalTrials.gov. NCT06704672, update 28 August 2026, checked 17 September 2026.

  14. [14]
    SmartGuide Predict pre/post behavior study (NCT07814508) · Trial registry · 2026-09-11

    ClinicalTrials.gov. NCT07814508, update 11 September 2026, checked 17 September 2026.