Conference Report

EASD 2025 — The Technologies That Will Reshape Diabetes Care

At EASD 2025 in Vienna, diabetes technology stopped behaving like a collection of separate gadgets and started looking like connected infrastructure. This is a summary of the key themes and what they mean going forward.

TL;DR — Key Points at a Glance

CGM becomes infrastructure: inpatient trials show +15% Time in Range and −24% complications; CGM variability analysis reveals five new diabetes subtypes.
AID scales across populations: toddlers, pregnancy, type 2 diabetes, and chronic kidney disease all show benefit; 24-month real-world durability confirmed.
Pregnancy head-to-head: CamAPS FX and Control-IQ outperform MiniMed 780G on neonatal outcomes and glycaemia.
Algorithm-first insulin therapy emerges: smart MDI and electronic glucose management systems (eGMS) push insulin decision-making beyond pumps.
Multi-analyte sensing arrives: Abbott’s dual glucose–ketone (DGK) sensor ecosystem enters pre-launch phase.
Regulation tightens: iCGM expectations rise; technology competency frameworks become essential — skills and training, not devices, are the bottleneck.
Trust becomes a feature: AID, CGM, and AI only work when psychological and clinical trust are deliberately designed into the system.

This content is for educational exploration only. It describes average responses and general principles. It is not medical advice and cannot replace individual clinical guidance from your diabetes care team.

What EASD 2025 Told Us About Diabetes Technology

At EASD 2025, diabetes technology stopped behaving like a set of separate gadgets and started to look like a connected infrastructure for care. Continuous glucose monitoring (CGM) is now operating like hospital telemetry. Automated insulin delivery (AID) systems are moving into almost every metabolic situation — toddlers, pregnancy, type 2 diabetes, and kidney disease. Algorithms are stepping outside pumps and into both hospital decision-support systems and smart injection regimens.

The message is simple: the biggest questions are no longer “does this work?” but “how do we scale this safely, fairly, and intelligently?” For people living with diabetes and the clinicians supporting them, this means more automation, more data, and more need for education, skills, and trust.

For an overview of how GNL thinks about CGM and technology, explore the CGM hub and the AID systems guide.

The Big Stories in One Read

EASD 2025 in Vienna marked a clear shift: rather than isolated device launches, there were whole-system moves — CGM in hospitals, AID across age groups and diagnoses, algorithmic insulin dosing in wards, and early signals of multi-analyte sensing.

On the CGM front, the DIETEC inpatient trial (n=166) showed that using CGM alongside structured titration and dedicated diabetes teams increased TIR by around 15 percentage points (63 to 78%) and reduced hospital complications by 24%, mainly infections. Patients on CGM were approaching 80–90% TIR by discharge, whilst those on traditional blood glucose monitoring plateaued around 60%. Accuracy remains strongest in non-ICU settings, but this is credible evidence that continuous glucose data can change hard outcomes in hospital.

Another CGM theme was classification. An observational study in 527 people used CGM variability profiles to define five metabolic subgroups: stable hyperglycaemia, postprandial spikes, brittle hypoglycaemia, dawn phenomenon, and erratic fluctuators. Each pattern maps onto different treatment strategies. CGM is no longer just a way to see glucose — it is becoming a way to re-draw the map of diabetes itself.

AID systems continued their expansion. MiniMed 780G secured CE-Mark coverage for children aged two years and older, pregnancy, insulin-requiring type 2 diabetes, and advanced CKD. The LENNY trial in toddlers (n=101) reported approximately a 10-point TIR gain and about a 0.6% HbA1c reduction, with parents reporting better sleep and less fear of hypoglycaemia. Longer-term data from Tandem’s Control-IQ show improvements at 12 months maintained at 24 months, and nationwide real-world data from France suggest AID benefits are durable at scale.

For pregnancy, a direct comparison between systems showed that CamAPS FX and Control-IQ delivered stronger neonatal and maternal outcomes compared with MiniMed 780G. This moves the field from “all systems are good” to a more nuanced view of which algorithms perform best in pregnancy physiology.

Smart MDI and electronic glucose management systems (eGMS) push towards algorithm-first insulin therapy, where the software decision layer matters as much as the delivery method. Abbott’s focus on ketone risk ahead of a dual glucose–ketone sensor signals the arrival of continuous metabolic monitoring beyond glucose alone.

Finally, rising regulatory expectations for iCGM, the University of Leicester’s four-level technology competency framework, and a consistent emphasis across sessions on “trust” all pointed in the same direction: technology is only as good as the team using it, and the people who trust it.

Theme Breakdown

1. Continuous Glucose Monitoring (CGM)

CGM in Hospitals

The DIETEC trial placed CGM at the centre of an inpatient type 2 diabetes protocol. With structured insulin titration and diabetes-specialist input, CGM users increased TIR from around 63% to 78%, whilst complication rates fell by 24%. Accuracy was strongest in non-ICU settings and more variable in ICU patients — critical care remains a frontier problem. The direction of travel is clear: for many inpatients, CGM can safely replace frequent point-of-care testing and deliver better outcomes.

Remote CGM Telemetry and EHR Integration

Speakers drew an explicit analogy between remote CGM telemetry and cardiac monitoring: nurses can watch multiple patients in real time and respond to alarms. The future lies not in new sensor chemistry, but in workflow and data integration — getting CGM data into hospital electronic records and building actionable, trusted alarm pathways.

CGM-Derived Metabolic Phenotypes

Using CGM variability metrics in 527 people, researchers identified five reproducible glucose profiles: stable hyperglycaemia, post-meal spikes, brittle hypoglycaemia, dawn phenomenon, and erratic fluctuators. Each phenotype points towards distinct therapeutic priorities — bolus optimisation, basal fine-tuning, hypo prevention, or a behavioural focus. This work nudges the field away from a one-size-fits-all HbA1c or TIR model towards pattern-based, CGM-informed subtypes.

Pregnancy and CGM

CGM continued to show benefits in pregnancy, including reduced risk of large-for-gestational-age infants and better maternal TIR. The meeting tone suggested CGM in pregnancy is moving from “recommended” towards “expected standard of care” in high-resource settings.

Beyond Glucose: Dual Glucose–Ketone (DGK) Sensing

Abbott used its Exhibit Hall presence to spotlight unmonitored ketones and trail its upcoming DGK sensor. For people prone to DKA or using treatments that modulate ketone metabolism, this could represent a major safety upgrade. This is less about a single product and more about a direction of travel: multi-analyte metabolic monitoring.

For a structured look at CGM choice and use, see Selecting a Continuous Glucose Monitor and Mastering CGM: 10 Top Lessons.

Automated Insulin Delivery (AID)

780G Across Toddlers, Pregnancy, T2D and CKD

MiniMed 780G secured CE-Mark coverage across four major groups: children aged two years and above, pregnancy, insulin-requiring type 2 diabetes, and advanced CKD. This reflects regulatory confidence that the algorithm can cope with very different physiologies and risk profiles — and underlines that AID is no longer a niche solution for the idealised adult type 1 diabetes user.

LENNY Trial: AID in Toddlers

In the LENNY study (n=101), MiniMed 780G increased TIR by about 10 percentage points and reduced HbA1c by around 0.6%, without new safety concerns. Parents reported better sleep and less fear of hypoglycaemia. From a lived-experience perspective, those last two outcomes are arguably as important as the clinical numbers.

Pregnancy: CamAPS FX and Control-IQ vs 780G

A key session directly compared AID systems in pregnant women with type 1 diabetes. CamAPS FX and Control-IQ both delivered superior maternal glycaemia and neonatal outcomes compared with 780G. This begins to establish a hierarchy of performance in pregnancy-specific physiology, moving beyond device-agnostic enthusiasm.

Durability and Real-World Population Data

Tandem reported that glycaemic improvements with Control-IQ at 12 months are maintained at 24 months. Nationwide French registry data showed durable AID benefits at scale. Together, these findings suggest that algorithm and behaviour interactions are stable over time in the real world, not just in trials.

AID Beyond the “Perfect User”

Several presentations highlighted benefits in people who do not regularly bolus or who live with complex co-morbidities. This challenges the assumption that AID is only for motivated, technology-savvy users and supports a more inclusive view of who can realistically benefit.

For help choosing a system, see the Choosing an AID System guide and the AID Systems series.

Algorithmic Insulin Therapy Beyond Pumps

Smart MDI

Medtronic showcased smart MDI — decision-support tools that help people using injections make more informed dosing decisions. Conceptually, this exports AID thinking into multiple daily injections: the algorithm starts to matter as much as the delivery method. For the many people worldwide who will never have access to pumps, this is potentially more disruptive than any new pump launch.

Electronic Glucose Management Systems (eGMS)

eGMS platforms apply computerised algorithms to insulin dosing in hospital wards, often combined with CGM data. Evidence is still mostly retrospective, but early results are promising. The key shift is that insulin decisions are becoming software-mediated inside hospitals, not just at home on personal devices.

Software-Hardware Decoupling

Smart MDI and eGMS together point towards a future where insulin therapy is governed primarily by software logic, and pumps, pens, or syringes are just delivery channels. This aligns with how dose-thinking and pattern recognition are explored in the Mealtime Insulin and Dynamic Glucose Management guides.

Systems, Regulation, and Human Factors

Rising iCGM Standards

Regulatory discussions highlighted evolving expectations for iCGM: higher accuracy thresholds, better interoperability, stronger data integrity, and clearer post-market surveillance. This will shape the next generation of Libre, Dexcom, and emerging sensors.

For context, see CGM Regulation and CGM Regulation: 2025 Update.

Competency Frameworks

The University of Leicester’s four-level technology competency framework treats CGM and AID as clinical skillsets that must be trained, assessed, and maintained. The true bottleneck is skills and training, not sensors and pumps. Expect more structured curricula, credentialing, and expectations around technology literacy for healthcare professionals.

Trust as a Core Design Parameter

Across multiple sessions, speakers returned to the idea of trust. It is not enough for algorithms to be accurate — people need to understand and predict how systems will behave. That includes clear alarms, transparent logic, robust safety nets, and realistic communication about limitations. Data must correspond with lived experience to be believed and used well.

What This Means Going Forward

For people with diabetes: CGM and AID are becoming more available and more capable, including in hospital and in pregnancy. The key challenge is choosing systems that fit individual life and learning to use them without burning out.
For clinicians: the priority is skills, not just prescribing. Interpreting CGM phenotypes, managing AID in complex populations, and understanding algorithm limitations are becoming core competencies.
For regulators and payers: the evidence base is now strong enough to justify broader access, but only if implementation is matched with training, support, and realistic expectations.
For industry: multi-analyte sensors, interoperable algorithms, and trustworthy user experience will matter more than marginal gains in accuracy.

Key Trials Referenced

DIETEC trial — Inpatient CGM vs blood glucose monitoring in type 2 diabetes (TIR and complications).
LENNY trial — MiniMed 780G in children aged 2–6 years.
Pregnancy AID trials — CamAPS FX and Control-IQ vs MiniMed 780G.
Control-IQ long-term real-world outcomes (12–24 months).
French nationwide AID registry — durability and population-level outcomes.
CGM variability phenotype study (five metabolic subgroups).