MiniMed Vivera (next-generation SmartGuard algorithm)
MiniMed
The right ambition. Almost none of the evidence yet.
What it is
MiniMed’s investigational algorithm is intended to make meal announcements optional. On 1 September 2026 the company reported US pivotal-trial enrollment complete and targeted US launch in the second half of 2027. No clearance or pivotal outcome results are established here; NMX8 registry studies do not explicitly use the Vivera brand name.
Editorial review: .
Most recent recorded citation date: 2026-09-02. 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
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.
- Benefit or performance
- Time in range: Provisional editorial assessment of the proposed product. No time-in-range results have been published for Vivera in type 1 diabetes at all. The feasibility study was 27 people over three months, and the registered trials are powered for *non-inferiority* in time in range (70–180 mg/dL / 3.9–10.0 mmol/L) against the system the person is already using — so the honest expectation is parity with less burden, not a higher ceiling.[2]
- Important harms and treatment burden
- Read the safety discussion and original sources. A missing summary does not establish safety.
- Approval and country access
- Investigational and not for sale. MiniMed’s September announcement supplies a company US launch target of second-half 2027; it does not establish an approved software update for every existing pump.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.
Research status alone does not establish approval, clinical benefit or local availability. 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.
Default calculation: 55 × 20 + 60 × 20 + 85 × 10 + 62 × 20 + 55 × 5 + 55 × 5 + 50 × 5 + 60 × 5 + 0 × 10 = 5490; divide by total weight 100. Unrounded weighted result: 54.9.
Provisional editorial assessment of the proposed product. No time-in-range results have been published for Vivera in type 1 diabetes at all. The feasibility study was 27 people over three months, and the registered trials are powered for *non-inferiority* in time in range (70–180 mg/dL / 3.9–10.0 mmol/L) against the system the person is already using — so the honest expectation is parity with less burden, not a higher ceiling.[2]
Provisional editorial assessment of the proposed product. Vivera inherits SmartGuard's predictive low management, which has a good track record on the 780G, but no Vivera-specific time-below-range or severe-hypoglycemia data has been published. An algorithm that doses for meals it detects itself has to be judged on its own lows, not its predecessor's.[1]
Provisional editorial assessment of the proposed product. The defining claim, and a real one: meal bolusing becomes optional, with automatic detection and dosing of unannounced meals. Scored below the fully-closed-loop research frontier because Vivera's version has only been shown in a 27-person feasibility study with no published outcomes.[1]
Provisional editorial assessment of the proposed product. Vivera is software, so its wearability is whatever it runs on: the tubed 780G and the tubed, screenless, phone-controlled MiniMed Flex. Control therefore depends on the host pump; Vivera itself adds meal automation, not phone control. The tubeless MiniMed Fit is a separate, later programme.[5]
Provisional editorial assessment of the proposed product. No mean glucose, GMI or HbA1c results have been published. The only public efficacy datum is a single anecdote — one 13-year-old reported going from 8% to 6.7% HbA1c — which is not evidence of anything at population level.[1]
Provisional editorial assessment of the proposed product. Machine-learning adaptation plus automatic meal dosing is exactly the recipe for flatter post-meal curves, and that is the point of the design — but it is untested at scale and nothing has been published.[1]
Provisional editorial assessment of the proposed product. No Vivera-specific exercise data. Unannounced activity is the hardest remaining problem for any algorithm that is also dosing for meals it detected itself, so this is scored cautiously until results exist.
Provisional editorial assessment of the proposed product. Three meal-handling modes — no bolus, a simple announcement, or a full carb count — is genuine user choice, and more than most systems offer. But it remains a closed MiniMed ecosystem with no access to the algorithm's internals.[1]
No current commercial access. Investigational; MiniMed targets US launch in second-half 2027 after completing US pivotal enrollment. No routine access or clearance established.[6]
AID default: 50% glucose outcomes and low-glucose protection, 40% daily experience, 10% access and cost. Freedom includes tubing, wearability, water-use limits and controller requirements. These are editorial priorities; studies and device generations differ. Check each scorecard for its evidence and limits.
Editor’s take
Making meal announcements optional could reduce daily work. The next evidence needed is the actual pivotal outcome report and labelled eligibility. Enrollment completion and a company launch target do not show how much glucose control, hypoglycemia or treatment burden will change.
The full picture
Most widely available insulin-only automated delivery systems remain hybrid: they adjust insulin automatically but ask users to announce meals. Vivera is MiniMed's answer to that — a third-generation control algorithm, using machine learning to adapt to the individual, in which the meal bolus becomes optional. You can let it detect the meal and dose for it, give it a simple "I'm eating" announcement, or carry on counting carbs exactly as you do now.1
It is software, not hardware. It is designed to run on pumps MiniMed already sells — the 780G and the newer screenless, phone-controlled MiniMed Flex — which is what makes it different from every other fully-closed-loop effort. Most of those are research systems reaching dozens of people. This one is aimed at people who already own those pumps.
What has actually been shown
Very little, and it is important to be blunt about that.
Vivera was unveiled on 12 March 2026 at the ATTD conference in Barcelona. The evidence behind the unveiling was a feasibility study of 14 adults and 13 children over three months. No time-in-range results have been published from it. No mean glucose, no HbA1c, no time-below-range. The single public efficacy datum in the coverage is an anecdote — one 13-year-old reported going from an HbA1c of 8% to 6.7%.1
You may see figures circulating in secondary coverage claiming roughly 74% time in range without meal announcement and 82% with. We have not been able to trace those to any primary source, so we do not publish them and neither should anyone else.
The trials — and what they are really testing
The registries are the clearest window on where this programme stands, with one important caveat: no registry record uses the name "Vivera". They all call the system the MiniMed NMX8 or the "Novel Medtronic Experimental AID System". The Vivera↔NMX8 link comes from press and manufacturer communication, not from ClinicalTrials.gov.
- GATEWAY (NCT07228117) — active follow-up. Active, not recruiting; enrollment began February 2026, 389 enrolled people across the US, Australia and New Zealand, randomised to no meal bolus, bolus at all meals, or bolus at will, for about 90 days. Its primary outcomes are descriptive — per-arm time in range (70–180 mg/dL / 3.9–10.0 mmol/L) and time below 70 mg/dL (3.9 mmol/L) — with primary completion in January 2027.4 It is registered against the NMX8 hardware that ships as MiniMed Flex, so it appears on this site under MiniMed Flex too. Its meal-bolus arms are the closest thing to a public test of the meal-optional idea.
- NEXUS (NCT07227805) — the experimental algorithm head-to-head against the 780G over 12 weeks, 116 people aged 2 and up with type 1 or type 2 diabetes. It began recruiting after an actual start on 19 July 2026, and now estimates completion on 17 March 2027.3
- ELEVATE (NCT07401901) — 230 adults with type 1 diabetes who are already on a commercial AID system with an HbA1c of 7.5% or above, across 21 sites in France, Germany, Italy, the Netherlands, Spain and the UK. It remained not yet recruiting on 8 August 2026, with a planned 15 October 2026 start, primary completion in December 2027, and final completion in March 2028.2
Read the endpoints carefully, because they set the honest expectation. Both NEXUS and ELEVATE test non-inferiority in time in range — the question they ask is "is Vivera no worse than the loop you already use?", not "is it better?". That is a defensible design: the thing being won here is the removal of carb counting, and matching your current control while doing far less work would be a genuine win. But it means nobody should expect a headline jump in time in range from this programme.
Why the ambition matters anyway
The reason meal announcement has survived is not laziness of design — it is the speed of injected insulin. Insulin under the skin peaks 1.5–2 hours after it is dosed, while a meal hits the bloodstream in 30–60 minutes. An algorithm that only reacts once it sees glucose rising is starting the race late. That is why the fully-closed-loop research literature reports meal-detection latencies of 25–40 minutes and post-meal excursions that automation alone cannot fully catch.
Vivera does not repeal that physics. What it does is make the compromise a choice rather than a requirement — bolus when you can be bothered, don't when you can't, and let the algorithm cover the gap. For a lot of people that trade will be worth several points of time in range.
What to watch
MiniMed reported US Vivera pivotal enrollment complete on 1 September 2026 and targets US launch in the second half of 2027. This is a manufacturer milestone, not published efficacy or clearance; the company release does not identify the pivotal study by NCT number. The NMX8 registry records therefore remain separately attributed.6
How to read this pipeline score. Design features and unreported outcomes are editorial projections. Trial entry ages are not an approved age indication. A study result applies only to its studied population and does not establish future commercial performance or access.
Coming soon
ETA · MiniMed targets US launch in the second half of 2027, subject to regulatory clearance; US pivotal enrollment was reported complete on 1 September 2026.
- →NEXUS (NCT07227805) — MiniMed's experimental algorithm versus the 780G over 12 weeks, n=116, non-inferiority in time in range · recruiting; started 19 July 2026, completion estimated 17 March 2027
- →ELEVATE (NCT07401901) — n=230 adults with type 1 diabetes already on a commercial AID system with HbA1c ≥7.5%, 21 sites across six European countries including the UK, non-inferiority in time in range versus their existing system · not yet recruiting; planned start October 2026, primary completion December 2027, study completion March 2028
- →GATEWAY (NCT07228117) — the study actually running: 389 people randomised to no meal bolus, bolus at all meals, or bolus at will, with descriptive per-arm outcomes · active, not recruiting; enrollment began February 2026; primary completion January 2027
- →US launch, dependent on results and regulatory clearance · Company target second half of 2027
Sources
- [1]MiniMed unveils Vivera: a next-generation fully-closed-loop algorithm · Science journalism · 2026-03-12 — Unveiling coverage from ATTD 2026. Feasibility study = 14 adults and 13 children over three months. No time-in-range figures are given anywhere in the piece; the only efficacy datum is a single anecdote (a 13-year-old going from 8% to 6.7% HbA1c).
MiniMed unveils Vivera: a next-generation fully-closed-loop algorithm — coverage of the ATTD 2026 unveiling, 12 March 2026. Feasibility study: 14 adults and 13 children, three months. No time-in-range figures reported.
- [2]ELEVATE — Evaluation of the Safety and Effectiveness of the Novel Medtronic Experimental AID System (NMX8) in Adults Living With Type 1 Diabetes · Trial registry — NCT07401901. Not yet recruiting as of 8 August 2026; planned start 15 October 2026, primary completion 4 December 2027 and study completion 29 March 2028. n=230, 21 sites across six European countries including the UK, adults with type 1 diabetes already on a commercial AID with HbA1c ≥7.5%. Primary endpoint is non-inferiority in time in range (70–180 mg/dL / 3.9–10.0 mmol/L) versus their existing system. The registry record does not use the name "Vivera".
ELEVATE — Evaluation of the Safety and Effectiveness of the Novel Medtronic Experimental AID System (NMX8) in Adults Living With Type 1 Diabetes. NCT07401901, not yet recruiting on 8 August 2026, planned start 15 October 2026, n=230, 21 sites across France, Germany, Italy, the Netherlands, Spain and the UK, primary completion 4 December 2027 and study completion 29 March 2028.
- [3]NEXUS — Evaluation of the Safety and Performance of the Novel Medtronic Experimental AID System (NMX8) in People Living With Diabetes · Trial registry — NCT07227805. Recruiting after an actual start on 19 July 2026; n=116, 12 weeks, versus the MiniMed 780G, enrolling people aged 2+ with type 1 or type 2 diabetes. Completion is estimated 17 March 2027. The registry record does not use the name "Vivera".
NEXUS — Evaluation of the Safety and Performance of the Novel Medtronic Experimental AID System (NMX8) in People Living With Diabetes. NCT07227805, recruiting after actual start 19 July 2026, n=116, versus the MiniMed 780G, completion estimated 17 March 2027; registry updated 29 July 2026.
- [4]GATEWAY — Safety Evaluation of the MiniMed NMX8-AID System in Children and Adults Living With Diabetes · Trial registry · 2026-09-02 — NCT07228117. Active, not recruiting; started 2 February 2026, actual n=389, US/Australia/New Zealand, primary completion January 2027. Randomises participants to no meal bolus, bolus at all meals, or bolus at will; primary outcomes are descriptive per-arm time in range and time below 70 mg/dL (3.9 mmol/L). The registry names the system NMX8-AID, not Vivera.
GATEWAY — Safety Evaluation of the MiniMed NMX8-AID System in Children and Adults Living With Diabetes. NCT07228117, active, not recruiting; began 2 February 2026, actual n=389, primary completion January 2027.
- [5]Closed-loop updates from ADA 2026 · Open-source community · 2026-06-18 — Community conference recap, not a MiniMed announcement. Sole source for the reported 2027 software update to the 780G and MiniMed Flex, so that timing is attributed rather than stated as fact.
- [6]MiniMed September 2026 pipeline and launch update · Manufacturer · 2026-09-01
MiniMed. September 2026 pipeline and commercial update.