CGM Glucose Prediction AI/ML Patent Landscape 2026
The CGM glucose prediction AI/ML space is dominated by a small cluster of medical-device incumbents and one academic foundation, with the top five filers accounting for 63% of the hundred largest filers’ combined output. The field reached a peak in 2021 and has since plateaued, signalling a maturing but still active competitive environment where differentiation increasingly depends on algorithm architecture and real-world integration.
Medtronic MiniMed leads a tightly concentrated field
Medtronic MiniMed holds the top position in the applicant ranking, followed closely by Dexcom and the University of Virginia. Patent Foundation — three entities that together account for a substantial majority of the leading filers’ activity.
The top five filers — Medtronic MiniMed, Dexcom, University of Virginia. Patent Foundation, WellDoc, and F. Hoffmann-La Roche — collectively hold 63% of the hundred largest filers’ combined patent records, indicating a pronounced tier gap between incumbents and the broader long tail of academic and startup applicants.
| # | Applicant | Patent records | Share |
|---|---|---|---|
| 1 | Medtronic MiniMed Inc | 107 | |
| 2 | Dexcom Inc | 93 | |
| 3 | University of Virginia Patent Foundation | 81 | |
| 4 | WellDoc Inc | 39 | |
| 5 | F HOFFMANN LA ROCHE & CO AG | 30 | |
| 6 | Ascensia Diabetes Care Holdings AG | 18 | |
| 7 | Abbott Diabetes Care Inc | 13 | |
| 8 | Roche Diabetes Care GmbH | 12 | |
| 9 | Insulet Corporation | 10 | |
| 10 | Willow Lab Inc | 9 |
| # | Applicant | Patent records | Share |
|---|---|---|---|
| 11 | University of Padova | 7 | |
| 12 | Roche Diabetes Care Inc | 6 | |
| 13 | Allez Health Inc | 5 | |
| 14 | IBM (International Business Machines Corporation) | 5 | |
| 15 | Frank Healthcare Advisors LLC | 5 | |
| 16 | Shanghai Sixth People’s Hospital | 5 | |
| 17 | Northeastern University (China) | 4 | |
| 18 | Tandem Diabetes Care Inc | 4 | |
| 19 | Marc D. Breton | 4 | |
| 20 | Tingo Medical Ltd | 4 |
The incumbents’ strong positions reflect deep integration of predictive algorithms into cleared CGM hardware ecosystems, creating a dual barrier of IP and regulatory precedent that later entrants must navigate.
Filing counts for 2025 and 2026 are likely under-reported due to standard patent publication lag of 18–24 months; apparent recent softness should not be read as a real decline. Longer-window growth, applicant concentration, and technology-route coverage are therefore more reliable signals than the latest-year bar alone.
Activity peaked in 2021; healthcare informatics dominates the technology mix
Annual filing volume climbed sharply to a 2021 peak, has eased since, and the technology composition is overwhelmingly concentrated in healthcare informatics and clinical diagnostics classes, with pure AI/ML computing classes present but secondary.
Annual filing trend
Filings rose from 35 in 2017 to a peak of 70 in 2021, then eased to the 43–57 range through 2024; 2025 and 2026 figures are artificially low due to publication lag and should not be interpreted as a continuing decline. The multi-year window still shows 11% growth over the period.
↗ Hover for values · click a bar to ask EurekaTechnology composition
G16H (Healthcare informatics) and A61B (Diagnosis & surgery) dominate the IPC mix, reflecting that prediction algorithms are filed primarily as medical-device claims. G06N (Computing based on AI models) is present but holds a smaller share, suggesting that pure ML methodology claims are less common than system-level clinical claims.
↗ Hover for values · click a bar to ask EurekaHighly cited patent families surfaced by the query
Citation-heavy patent families returned by the query. Use this section as citation context, not as a curated list of the most topic-specific patents.
Accuracy continuous glucose monitoring method, sys…
A method, system, and device for improving the accuracy of a continuous glucose monitoring sensor by estimating a CGM signal at a time t+PH using a value of CGM at time t, using a real-time short-time glucose prediction horizon to estimate the real time denoised CGM value with a noise estimation algorithm. (excerpt from the patent abstract)
| # | Patent | Citations |
|---|---|---|
| 1 | Medical device for predicting a user’s future glyc… | 205 |
| 2 | Systems and methods for patient monitoring using a… | 172 |
| 3 | Medical device for predicting a user’s future glyc… | 162 |
| 4 | CGM-based prevention of hypoglycemia via hypoglyce… | 152 |
| 5 | Electrochemical glucose sensing by equilibrium glu… | 151 |
| 6 | CGM-based prevention of hypoglycemia via hypoglyce… | 138 |
| 7 | Method, system and computer program product for re… | 108 |
| 8 | CGM-Based Prevention of Hypoglycemia Via Hypoglyce… | 82 |
Ranked by total forward citations. Citation counts favour older and broadly cited patent families, and broad or adjacent patents may appear when they match the search scope. Treat this section as citation context, not as a curated list of the most topic-specific patents. Some patent titles may be shown in their original, non-English language where an accurate translation could not be guaranteed.
What the competitive structure means for R&D investment
The combination of a maturing lifecycle, high incumbent concentration, active cross-sector collaboration, and US-centric jurisdiction coverage shapes where meaningful R&D differentiation remains available.
Field has plateaued near its 2021 peak
The lifecycle stage is Maturity: annual filings have plateaued near their 2021 peak, though the multi-year window still shows positive growth. New entrants should expect crowded claim space in core prediction architectures and should target differentiated algorithm designs — such as personalized patient-state models or multimodal input fusion — rather than broad system claims already staked by incumbents.
Lifecycle: Mature / PlateauedTop five filers dominate; long tail is fragmented
The top five filers hold 63% of the hundred largest filers’ combined patent records, and the gap to the sixth-ranked applicant (Ascensia Diabetes Care) is substantial. This concentration means freedom-to-operate searches must prioritise the Medtronic MiniMed, Dexcom, University of Virginia, WellDoc, and Roche portfolios before any product development. The long tail of academic and individual inventors suggests claim fragmentation in experimental sub-fields.
High concentrationRoche entities and Dexcom–Padova are the most active co-filers
The most active co-filing pairs are Dexcom with the University of Padova (6 joint records), F. Hoffmann-La Roche with Roche Diabetes Care GmbH (6 records), and F. Hoffmann-La Roche with IBM (3 records). These relationships signal that sensor-algorithm integration and AI-platform licensing are active collaboration vectors. The University of Virginia Patent Foundation co-filed with Abbott Diabetes Care (2 records), illustrating how academic IP can feed directly into commercial pipelines.
Cross-sector collaborationUS-centric filings; Europe and PCT are secondary priorities
The United States is the lead jurisdiction by a wide margin, followed by Europe (EPO) and WIPO PCT filings. Canada, China, Australia, and India make up a secondary tier. Japan and other Asian markets show markedly lower coverage, which may reflect either strategic prioritisation by incumbents or a gap that regional players could exploit. Any global product launch strategy should treat US prosecution as the primary battleground.
US-led geographyGo beyond the landscape: Eureka’s TRIZ Solution agent breaks down an R&D problem and returns patented concept solutions, each with a technical approach and cited patent & literature evidence.
| Applicant | Collaborator | Co-filings |
|---|---|---|
| Dexcom Inc | University of Padova | 6 |
| F. Hoffmann-La Roche Ltd | Roche Diabetes Care GmbH | 6 |
| F. Hoffmann-La Roche Ltd | IBM (International Business Machines Corporation) | 3 |
| Roche Diabetes Care GmbH | Roche Diabetes Care Inc | 3 |
| University of Virginia Patent Foundation | Abbott Diabetes Care Inc | 2 |
| F. Hoffmann-La Roche Ltd | Roche Diabetes Care Inc | 2 |
| Roche Diabetes Care GmbH | IBM (International Business Machines Corporation) | 2 |
Co-filing pairs, ranked by the number of jointly-filed patent families.
Medtronic MiniMed and Dexcom lead; University of Virginia accelerating
The top two positions are held by integrated device makers whose portfolios span sensor hardware, data processing, and prediction algorithms; the University of Virginia. Patent Foundation is the leading academic filer and has shown notably stronger recent momentum.
Medtronic MiniMed Inc
Medtronic MiniMed holds 107 patent records — the largest portfolio in the ranking — with a technology emphasis concentrated in A61B5 (diagnosis and surgery), G01N27 (electrochemical material analysis), and A61M5 (drug delivery devices), reflecting its closed-loop insulin delivery heritage. Recent momentum shows a 14% decline versus the prior three-year period, consistent with a maturing and rationalising portfolio rather than an exit from the space.
patent records: 107Dexcom Inc
Dexcom holds 93 patent records, with its technology emphasis concentrated in G16H50 (healthcare informatics — decision support), A61B5, and G16H40 (healthcare informatics — management), indicating a software and data-platform orientation relative to Medtronic MiniMed’s hardware focus. Recent momentum shows a 24% decline versus the prior period, but Dexcom’s collaboration with the University of Padova (6 joint records) suggests continued investment in academic partnerships for algorithmic development.
patent records: 93| Applicant | Recent (3 yrs) | Trend |
|---|---|---|
| Medtronic MiniMed Inc | 25 | ▼ -14% |
| Dexcom Inc | 29 | ▼ -24% |
| University of Virginia Patent Foundation | 23 | ▲ +77% |
| WellDoc Inc | 18 | ▬ 0% |
| F. Hoffmann-La Roche Ltd | 3 | ▲ new entrant |
| Abbott Diabetes Care Inc | 3 | ▲ new entrant |
| Roche Diabetes Care GmbH | 3 | ▲ new entrant |
| Insulet Corporation | 10 | ▲ new entrant |
Under-served adjacent branches worth monitoring
Several IPC branches present in the corpus hold materially lower filing shares relative to the dominant healthcare informatics and diagnostics classes; two of these have plausible technical value and realistic entry paths for teams already working in predictive CGM.
G06N · Computing based on AI models
With 82 patent records and a 7% share of the corpus, pure AI/ML computing claims are sparse relative to the volume of clinical system claims. This gap suggests that foundational machine-learning methodology patents — covering novel neural architectures, uncertainty quantification, or federated learning approaches for glucose prediction — may be less contested than system-level claims. Teams with strong ML research capability could file method claims here without immediately conflicting with the dominant device-system portfolios held by Medtronic MiniMed and Dexcom.
Search this in Eureka →G06F · Electric digital data processing
G06F covers data processing infrastructure including real-time signal processing, software architecture, and interoperability standards — present at 54 patent records and 5% share. As CGM platforms increasingly integrate with electronic health records, wearables, and cloud-based decision engines, software architecture and data-pipeline patents in this branch represent an under-served adjacent area. Entrants with expertise in embedded systems or health data interoperability (e.g. FHIR integration) could find relatively open claim space here.
Search this in Eureka →How leaders differ by technology route
Route coverage across the main technology branches in the current evidence set.
| Player | A61B 5 · Diagnosis & surgery | G16H 50 · Healthcare informatics | G16H 20 · Healthcare informatics | G16H 40 · Healthcare informatics | A61M 5 · Devices for body fluids |
|---|---|---|---|---|---|
| University of Virginia Patent Foundation | Strong · 56 | Strong · 78 | Strong · 46 | Strong · 46 | Moderate · 27 |
| Dexcom Inc | Strong · 64 | Strong · 84 | Moderate · 32 | Strong · 46 | Emerging · 10 |
| Medtronic MiniMed Inc | Strong · 104 | Moderate · 27 | Moderate · 22 | Moderate · 27 | Moderate · 31 |
| WellDoc Inc | Strong · 23 | Strong · 39 | Strong · 26 | Absent | Absent |
| F. Hoffmann-La Roche Ltd | Moderate · 8 | Strong · 28 | Strong · 23 | Strong · 17 | Moderate · 7 |
| Abbott Diabetes Care Inc | Strong · 8 | Strong · 11 | Strong · 10 | Strong · 9 | Strong · 7 |
| Roche Diabetes Care GmbH | Absent | Strong · 11 | Strong · 10 | Strong · 9 | Absent |
Frequently asked questions
Medtronic MiniMed Inc leads the applicant ranking with 107 patent records, ahead of Dexcom Inc (93) and the University of Virginia Patent Foundation (81).
The top five filers — Medtronic MiniMed, Dexcom, University of Virginia Patent Foundation, WellDoc, and F. Hoffmann-La Roche — account for 63% of the hundred largest filers’ combined patent records, indicating a highly concentrated competitive landscape.
The field is assessed as Mature: annual filings peaked in 2021 and have since plateaued. The multi-year window still shows 11% growth, but annual volume has eased from its peak, consistent with a field entering consolidation rather than early-stage expansion.
The United States is the lead jurisdiction by a wide margin, with Europe (EPO) and WIPO PCT filings forming a secondary tier. Canada, China, Australia, and India hold smaller but meaningful coverage; Japan and other Asian markets show relatively low filing density.
The most active co-filing pairs are Dexcom with the University of Padova (6 joint records) and F. Hoffmann-La Roche with Roche Diabetes Care GmbH (6 records). F. Hoffmann-La Roche also co-filed with IBM on 3 records, and the University of Virginia Patent Foundation co-filed with Abbott Diabetes Care on 2 records.
The G06N (Computing based on AI models) and G06F (Electric digital data processing) branches hold lower filing shares — 7% and 5% respectively — relative to the dominant healthcare informatics and diagnostics classes. These branches represent adjacent areas where foundational ML methodology and data-infrastructure claims may be less contested, though they should be validated with a full freedom-to-operate search before filing.
Ready to map your own CGM AI/ML patent landscape?
Join 18,000+ innovators using PatSnap Eureka to map any technology landscape: search 2B+ patents and papers, surface key assignees, and generate a report like this in minutes.
Disclaimer. This page is generated from PatSnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. PatSnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.