Skip to content
type1.science

CamAPS FX

CamDiab (Cambridge, UK) / distributed by Ypsomed as mylife CamAPS FX

A phone-based hybrid closed-loop app running the adaptive Cambridge model-predictive-control algorithm. Uniquely licensed from age 1 and through pregnancy, it pairs with Dexcom G6, Dexcom G7 (live from June 2026) and FreeStyle Libre 3 sensors driving the YpsoPump (and DANA pumps) — the broadest eligibility of any commercial system, backed by two NEJM randomized trials. UK/EU; not FDA-cleared.

Available nowStrong evidencehybrid-closed-loopphone-controltubedOfficial site ↗

The scorecard

Time in range74

Achieved real-world TIR of 69.6% (median, IQR 61.3-77.2) across 35,714 T1D users in 19 countries (Boughton et al., Metabologia 2026) is the most representative figure; a smaller 1,805-user real-world analysis reported a higher 72.6% ± 11.5%. By age, real-world TIR spans ~62.5% (young adults 18-22y) to ~76.1% (≥65y). Weighting toward the large, broad, multi-age cohort at ~70% places CamAPS FX at 74 on the achieved-TIR scale calibrated to the MiniMed 780G anchor (~76% = 80) — genuinely below the 780G anchor, kept conservative because the dominant dataset sits at ~70% rather than the smaller cohort's ~73%.[1]

Hypo protection76

Achieved real-world time-below-70 mg/dL (<3.9 mmol/L) was 2.5% (median, IQR 1.4-3.9) in the 35,714-user cohort and 2.3% (IQR 1.3-3.6) in the 1,805-user cohort, staying <4% across every age group. A TBR of ~2.3-2.5% maps to the 76-78 band (the calibration places ~2-2.5% at 76-78, ~3% at 72-74); CamAPS FX's well-controlled hypoglycemia at scale supports 76.[1]

Automation level70

Adaptive algorithm auto-titrates basal every 8-12 min and self-learns daily/meal patterns, but it remains hybrid: meals are announced. The Cambridge algorithm is unusually forgiving of imprecise meal input — the randomized SMASH trial found simplified meal announcement (picking a meal "size") non-inferior to full carb counting for time-in-range (69.9% vs 70.7%), and a smaller 6-month study in 34 adults reported at ADA 2026 pointed the same way (65% vs 70% time-in-range with loose meal estimation versus precise counting, no statistically significant difference). Still a hybrid loop, so the score is unchanged: meals must be announced.

Average glucose74

Achieved GMI was 7.1% ± 0.5% (mean glucose 8.7 mmol/L) in the 35,714-user real-world cohort and 6.9% (mean glucose 8.4 mmol/L) in the 1,805-user cohort. A GMI near 7.1% reflects solid but not class-leading average-glucose control, consistent with the ~70% achieved TIR; scored 74 to align with the time-in-range anchor.[1]

Low variability73

Glycemic variability is not reported as a coefficient of variation in the primary real-world dataset, but the consistent TBR (~2.5%) alongside a ~70% TIR and GMI 7.1% across a 35,714-user, all-ages cohort indicates good day-to-day stability without exceptional tightening. Scored 73, slightly below the TIR anchor to reflect the absence of a directly reported low-CV figure and the conservative read of the larger cohort.[1]

Exercise handling74

Dedicated Ease-off (raises target, cuts insulin for activity) and Boost (raises insulin) modes; real-world analysis of 7,464 users found Ease-off lowered time above range with no safety signal.

Customizability84

Fully adjustable personal glucose target 4.4-11.0 mmol/L (default 5.8) by time of day, plus Boost/Ease-off and slowly-absorbed-meal options — among the most tunable commercial systems. Hardware choice widened again in June 2026 when Dexcom G7 joined Dexcom G6 and FreeStyle Libre 3 / 3 Plus as a supported sensor, on top of a choice of pumps (YpsoPump, DANA).[5]

Access & cost55

Available across UK/EU (and Australia/others) on Android and, since 2025, iOS; reimbursed in several systems. Sensor choice broadened in June 2026 — Dexcom G7 went live in Switzerland on 3 June with a rollout to 16 further countries at the end of June, subject to local approvals — which makes the system easier to fit around whichever CGM you can actually get. But it is still not FDA-cleared, so it remains unavailable in the US, and the score is unchanged.[5]

Freedom (form factor)55

Runs on small tubed pumps (YpsoPump, Dana) — compact and discreet for a tubed system, but not a tubeless patch.

Glycemic criteria are scored on the levels actually achieved in large real-world Type 1 diabetes cohorts — not the headline improvement over a trial's baseline (an improvement that looks bigger when the starting population was doing poorly). Type 2 diabetes trial data is never used to score a Type 1 system; where only improvement data exists, it informs the rationale, not the score. Freedom captures form factor and wearability, so a tubeless system is rewarded for the mobility a tubed one can't match.

The full picture

CamAPS FX is a hybrid closed-loop app for type 1 diabetes that runs the adaptive "Cambridge" model-predictive-control algorithm on a smartphone, where it links a continuous glucose monitor (CGM) to an insulin pump.1 Rather than being a single sealed device, it is the algorithm and app from CamDiab that drives compatible hardware: the mylife YpsoPump (a small tubed patch-style pump) and DANA Diabecare RS / DANA-i pumps, paired with Dexcom G6, Dexcom G7 or Abbott FreeStyle Libre 3 / Libre 3 Plus sensors.12 The algorithm recalculates and adjusts insulin every 8-12 minutes from CGM readings, and continuously "learns" the user's daily, day-to-day and around-meal insulin needs.3

New: Dexcom G7 support (June 2026). Dexcom G7 is no longer a promise — it went live in the mylife Loop combination (YpsoPump + CamAPS FX) in Switzerland on 3 June 2026, with rollout announced for the end of June 2026 across Austria, Belgium, Czech Republic, Denmark, Finland, France, Germany, Ireland, Italy, the Netherlands, New Zealand, Norway, Poland, Spain, Sweden, the UK and Australia — in each case subject to local regulatory approval, so the date you can actually switch depends on where you live.4 In practice this means you can now run the system on Dexcom G6, Dexcom G7 or FreeStyle Libre 3 / 3 Plus, which is the widest sensor choice of any commercial closed loop and makes it far likelier you can build a loop around whichever CGM your health system will fund.

What the trials show. Its evidence base is unusually strong — two New England Journal of Medicine randomized trials. In very young children aged 1-7 years (KidsAP02, NCT03784027), CamAPS FX raised time-in-range (70-180 mg/dL) by 8.7 percentage points versus a sensor-augmented pump, cut mean glucose by 12.3 mg/dL and HbA1c by 0.4 points, with no increase in hypoglycemia.5 In pregnancy (AiDAPT), it lifted time in the tight pregnancy target (63-140 mg/dL) to 68.2% versus 55.6% on standard care — a 10.5-point gain — and reduced hyperglycemia.6 Benefits persisted through the first 6 months postpartum (72% vs 54% time-in-range).7 Real-world data from 1,805 users mirror the trials: average time-in-range 72.6% (rising to ~82% in users over 65) and a glucose management indicator (GMI, an estimated HbA1c) of 6.9%.8

Does it hold up over years, not months? A Cambridge paper published on 1 July 2026 followed 64 children aged 1-7 through an 18-month home extension after the trial. HbA1c fell from 7.3% at baseline to 6.8% during the trial phase and was still 6.8% at the end of the extension — the benefit held rather than drifting back, which is the question that matters once the novelty and the study-team support wear off.9 One honest caveat: that study's actual purpose was to measure parents' expectations of closed-loop over long-term home use (its primary outcome is a parental-expectancy questionnaire), and the HbA1c figures are a secondary observation from the same cohort — so read them as supporting evidence for durability, not as a fresh glycemic-outcomes trial.

How automated is it? It is a hybrid loop: it automates background (basal) insulin and adds automatic correction doses, but you still announce meals and enter carbohydrates so it can deliver a mealtime bolus.3 Encouragingly, a randomized trial (SMASH) found that simplified meal announcement — picking a meal "size" instead of counting carbs precisely — gave non-inferior time-in-range (69.9% vs 70.7%), easing one of the most tedious daily tasks.10 A smaller 6-month study in 34 adults, reported at ADA 2026, pointed in the same direction: loose meal estimation reached 65% time-in-range against 70% with precise carb counting, with no statistically significant difference between the groups.11 Treat that one lightly — it is a conference report rather than a published paper, and SMASH remains the stronger evidence — but the two together suggest the adaptive Cambridge algorithm is unusually forgiving of imprecise meal input. That is the practical middle ground between a strict hybrid loop and a fully closed one.

Exercise and customization. Two manual modes handle changing needs: Ease-off raises the glucose target and trims insulin ahead of exercise or other low-risk periods, and Boost increases insulin for illness or stubborn highs. A real-world analysis of 7,464 users found Ease-off reduced time above range and Boost was used safely, with no severe hypoglycemia or DKA attributed to either feature.12 Users can set a personal glucose target anywhere from 4.4 to 11.0 mmol/L (default 5.8 mmol/L), adjustable by time of day — among the most tunable commercial systems.13

Ages, indications and access. CamAPS FX is licensed for people with type 1 diabetes aged 1 year and older and is the only system specifically indicated for use in pregnancy, with a wide approved range of 5-350 units/day total daily insulin.12 (Minimum approved age varies slightly by sensor.) It carries a CE mark and is available across the UK, EU, Australia and some other markets — historically Android-only, with an iOS/iPhone version launched in 2025.214 It is not FDA-cleared, so it is not available in the US.

What's coming. The 2025 iOS launch (rolling out country by country) and a new mylife Cloud data-sharing function are expanding reach and remote monitoring.14 The Dexcom G7 integration has now shipped and is working through the remaining countries as local approvals land,4 while further sensor integrations and additional pump partners continue to widen the choice of hardware.1 The same Cambridge algorithm is being studied in fully closed-loop, no-meal-announcement configurations and in type 2 diabetes — the likely next frontiers for the platform.56 The one thing still not on the horizon is the US: there is no FDA clearance, and none has been announced.

References

  1. CamDiab. CamAPS FX — closed-loop insulin delivery system for type 1 diabetes (compatible pumps, sensors, ages, pregnancy). https://camdiab.com/ ↩ ↩2 ↩3 ↩4

  2. Ypsomed. Ypsomed and CamDiab expand availability of mylife CamAPS FX on iOS (compatible YpsoPump + Dexcom G6 and FreeStyle Libre 3/3 Plus; ages; pregnancy), 10 June 2025. https://www.ypsomed.com/en/news-insights/news/press-releases/news-reader-detail-page/ypsomed-and-camdiab-expand-availability-of-mylife-camaps-fx-on-ios ↩ ↩2 ↩3

  3. CamDiab / Health Innovation Network. CamAPS FX closed-loop system specifications (adaptive model predictive control; adjusts every 8-12 min; default target 5.8 mmol/L). https://healthinnovationnetwork.com/wp-content/uploads/2021/11/CamAPS_FX_-Specifications.pdf ↩ ↩2

  4. Drug Delivery Business. mylife (Ypsomed) / CamDiab integrate Dexcom G7 into mylife Loop — live in Switzerland 3 June 2026, 16 further countries from end of June 2026, subject to local regulatory approval, 3 June 2026. https://www.drugdeliverybusiness.com/mylife-integrates-myloop-pump-dexcom-g7/ ↩ ↩2

  5. Ware J, Allen JM, Boughton CK, et al. Randomized Trial of Closed-Loop Control in Very Young Children with Type 1 Diabetes. N Engl J Med 2022;386:209-219. doi:10.1056/NEJMoa2111673. https://pubmed.ncbi.nlm.nih.gov/35045227/ ↩ ↩2

  6. Lee TTM, Collett C, Bergford S, et al. Automated Insulin Delivery in Women with Pregnancy Complicated by Type 1 Diabetes (AiDAPT). N Engl J Med 2023;389:1566-1578. doi:10.1056/NEJMoa2303911. https://pubmed.ncbi.nlm.nih.gov/37796241/ ↩ ↩2

  7. Lee TTM, Collett C, Bergford S, et al. Automated insulin delivery during the first 6 months postpartum (AiDAPT extension). Lancet Diabetes Endocrinol 2025;13:210-220. doi:10.1016/S2213-8587(24)00340-1. https://pubmed.ncbi.nlm.nih.gov/39884300/ ↩

  8. Alwan H, Wilinska ME, Ruan Y, Da Silva J, Hovorka R. Real-World Evidence Analysis of a Hybrid Closed-Loop System (mylife CamAPS FX). J Diabetes Sci Technol 2025;19:385-389. doi:10.1177/19322968231185348. https://pmc.ncbi.nlm.nih.gov/articles/PMC11873883/ ↩

  9. Vidou A, et al. Cambridge hybrid closed-loop use in very young children with type 1 diabetes: parental expectancies during long-term home use. Diabet Med 2026 (published 1 July 2026). doi:10.1111/dme.70408. (n=64, ages 1-7; HbA1c 7.3% at baseline → 6.8% in the trial phase → 6.8% after an 18-month home extension. Primary outcome was the parental-expectancy questionnaire; HbA1c is secondary.) https://doi.org/10.1111/dme.70408 ↩

  10. SMASH investigators. Simplified meal announcement study (SMASH) using hybrid closed-loop insulin delivery in youth and young adults with type 1 diabetes. Diabetologia 2024. doi:10.1007/s00125-024-06319-w. https://pubmed.ncbi.nlm.nih.gov/39560745/ ↩

  11. Closed-loop updates from ADA 2026 — conference recap covering a 6-month CamAPS FX / mylife Loop study (n=34 adults) comparing loose meal estimation (65% time-in-range) with precise carb counting (70%), no statistically significant difference. Conference recap, not a primary published abstract. https://www.diabetotech.com/blog/closed-loop-updates-from-ada-2026 ↩

  12. Royston C, Bergford S, Calhoun P, et al. Safety of Options to "Boost" and "Ease-Off" in the CamAPS FX Hybrid Closed-Loop System: A Real-World Analysis. Diabetes Technol Ther 2025;27:60-65. doi:10.1089/dia.2024.0298. https://pmc.ncbi.nlm.nih.gov/articles/PMC7617696/ ↩

  13. Evaluating the Impact of Applying Personal Glucose Targets in a Closed-Loop System (CamAPS FX; default 5.8 mmol/L, adjustable 4.4-11.0 mmol/L). J Diabetes Sci Technol 2022. doi:10.1177/19322968221145184. https://pubmed.ncbi.nlm.nih.gov/36540007/ ↩

  14. Ypsomed. mylife CamAPS FX on iOS — 2025 expansion and mylife Cloud data sharing. https://www.ypsomed.com/en/news-insights/news/press-releases/news-reader-detail-page/ypsomed-and-camdiab-expand-availability-of-mylife-camaps-fx-on-ios ↩ ↩2

What's next for this

  • →iOS/iPhone version rolling out country by country · 2025
  • →New mylife Cloud data-sharing function for remote monitoring
  • →Dexcom G7 rollout completing across the remaining mylife Loop countries, where local approvals allow · 2026
  • →Further sensor integrations and additional pump partners
  • →Cambridge algorithm being studied in fully closed-loop (no meal announcement) and type 2 diabetes configurations

Sources

  1. [1]Real-world evidence on the CamAPS FX hybrid closed-loop system in people living with type 1 diabetes · peer-reviewed · 2026-01-01
  2. [2]Real-World Evidence Analysis of a Hybrid Closed-Loop System (CamAPS FX, 1,805 users / 15 countries) · peer-reviewed · 2023-01-01
  3. [3]AiDAPT: Closed-Loop Insulin Delivery in Pregnant Women with Type 1 Diabetes (NEJM) · peer-reviewed · 2023-01-01
  4. [4]Long-term assessment of the NHS hybrid closed-loop real-world study in children and young people with type 1 diabetes (incl. CamAPS FX) · peer-reviewed · 2024-01-01
  5. [5]mylife (Ypsomed) / CamDiab integrate Dexcom G7 into mylife Loop — Switzerland from 3 June 2026, 16 further countries from end of June 2026 · news · 2026-06-03
  6. [6]Vidou A, et al. Cambridge hybrid closed-loop use in very young children with type 1 diabetes — parental expectancies during long-term home use (Diabet Med, 2026) · peer-reviewed · 2026-07-01 — Primary outcome is the INSPIRE parental-expectancy questionnaire; the HbA1c figures quoted here are a secondary observation.
  7. [7]Closed-loop updates from ADA 2026 (conference recap) · community — Conference recap, not a primary abstract — the ADA 2026 meal-estimation numbers are reported on this basis and are not used to move any score.