Diabetes Management Technology

Scope and core evidence base

This deep-dive covers three linked technology classes:

1. Continuous glucose monitors (CGMs) — sensors that provide interstitial glucose readings and trend data.
2. Insulin pumps — durable electromechanical delivery devices.
3. Automated insulin delivery (AID)/closed-loop systems — sensor, pump, and control algorithm operating together.

Core sources used for load-bearing claims:

Technology architecture and technical claims

2.1 Continuous glucose monitors

A CGM uses a sensor inserted under the skin to measure glucose continuously and transmit readings to a display device or phone. The FDA’s description of Dexcom’s Stelo states that the device is a continuous glucose monitor with a small sensor placed just under the skin that provides glucose readings without a prescription S6: FDA press release, 2024-03-05, [fda.gov .

CGM devices are distinct from traditional blood-glucose meters. ISO 15197:2013 specifies accuracy requirements for blood-glucose monitoring systems for self-testing, not for CGM sensors S5: ISO 15197:2013, 2013-05, [iso.org . Therefore, a claim that “ISO 15197 applies to CGM” would be a category error; the standard is for SMBG, while CGM performance is typically evaluated with different metrics such as mean absolute relative difference (MARD) and time-in-range metrics. This is an important source-bound distinction.

2.2 Insulin pumps and automated insulin delivery

The foundational granted patent here is US 6,558,351 B1, assigned to Medtronic MiniMed, Inc., granted 2003-05-06, titled Closed loop system for controlling insulin infusion S1: US 6,558,351 B1, granted 2003-05-06, [patents.google.com . The patent describes a closed-loop system that uses a sensor to measure glucose and a controller to regulate insulin infusion — the basic architecture that later commercial AID systems build upon.

This is a granted patent, not a pre-grant publication or application. That matters because only a granted patent can support a claim of recorded IP rights; however, a patent is evidence of a claimed invention, not proof of clinical efficacy, market adoption, or commercial viability. The grant date alone shows the broad closed-loop concept has been in the IP record since 2003.

2.3 Clinical evidence for AID

The strongest peer-reviewed evidence cited here is Brown et al., Six-Month Randomized, Multicenter Trial of Closed-Loop Control in Type 1 Diabetes, New England Journal of Medicine, DOI 10.1056/NEJMoa1907863, published 2019-10-31 S2: [nejm.org . This RCT compared a closed-loop AID system against a sensor-augmented pump in patients with type 1 diabetes. The primary outcome was time in the target glucose range of 70–180 mg/dL.

The trial reported that the closed-loop group had a higher time-in-range than the control group: 71% vs. 59%, a difference of about 11 percentage points [S2: Brown et al., DOI 10.1056/NEJMoa1907863, 2019-10-31]. The trial also reported improvement in HbA1c; exact effect sizes should be read from the paper, but the directional finding is well documented in the source.

2.4 Clinical metrics: time-in-range and estimated A1c

Battelino et al. provide an international consensus on CGM metrics: time-in-range 70–180 mg/dL >70%, time below range <4%, time above range <25%, and a goal of minimizing clinically meaningful hypoglycemia S3: Battelino et al., DOI 10.2337/dci19-0028, Diabetes Care 2019-08-01, [doi.org .

The ADA Standards of Care 2024 provide the estimated average glucose (eAG) formula:

eAG (mg/dL) = 28.7 × A1C − 46.7

This is used to translate a measured A1C into an average glucose or, inversely, to estimate A1C from mean glucose S4: ADA Standards, DOI 10.2337/dc24-S006, 2024-01-01, [doi.org .

Intellectual property and source-bound claims

US 6,558,351 B1 is a granted patent. It supports the claim that a closed-loop sensor/controller/infusion system was recorded in the patent system in 2003 [S1]. It does not support the claim that any specific modern CGM or AID product is protected by this patent alone. Modern commercial systems are covered by later patent families, and those later patents were not retrieved for this note. Therefore, any specific modern product IP claim should be marked as needs primary source validation unless a current granted patent is retrieved and verified.

Likewise, the RCT result from Brown et al. [S2] supports the claim that closed-loop AID can improve time-in-range in a randomized trial population. It does not by itself prove that every commercially marketed AID system performs identically, nor does it prove real-world effectiveness across all users, payers, or age groups.

For company-specific performance claims, investor and marketing materials are Tier 2 self-reported unless independently verified by an FDA summary or peer-reviewed study. For example, DexCom’s 10-K risk factors discuss regulatory, reimbursement, competitive, and adoption risks S7: DexCom 10-K listing, EDGAR CIK 0001093557, accessed 2026-04-27, [sec.gov . The existence of those risk factors is a primary-source company disclosure; the risk factors themselves are not independent proof of device performance.

Unit economics / business model

The diabetes-management technology value chain monetizes through a combination of:

• Durable hardware — insulin pumps and reusable transmitters sold to patients or DME suppliers.
• Recurring disposable revenue — CGM sensors replaced every 7–14 days, pump infusion sets and reservoirs replaced every 2–3 days.
• Software/algorithm value — embedded in devices and reflected in system pricing; no standalone software-only subscription model is confirmed from the sources retrieved here.
• OTC cash-pay channel — FDA cleared Dexcom’s Stelo as an over-the-counter CGM, creating a non-prescription, consumer-pay product for people not on insulin [S6: FDA press release, 2024-03-05].

Who captures value:

• Sensor makers capture recurring sensor revenue; recurring disposables are structurally higher-margin than one-time hardware, but the specific gross margin for any company must be labeled needs primary source validation unless read directly from a 10-K line item or audited financial statement.
• Pump makers capture durable pump revenue plus disposable infusion-set revenue.
• Insulin manufacturers capture insulin revenue, and connected-pen/glucose-data integrations may strengthen their ecosystem positions, but that connection is inferred from company strategy and was not independently quantified in this note.
• Payers and health systems may capture downstream savings from avoided complications, but the magnitude is derived/inferred and not measured in the sources retrieved here.

The precise prices for Dexcom G7 sensors, Abbott FreeStyle Libre sensors, or Medtronic pump systems are not stated here because no primary-source pricing document was retrieved. Any ASP/margin figure cited from memory would be fabrication and is therefore omitted.

Quantitative worked calculation

8.1 Formula

Use the ADA estimated average glucose formula [S4: ADA Standards of Care 2024, DOI 10.2337/dc24-S006, 2024-01-01]:

\[

\text{eAG (mg/dL)} = 28.7 \times \text{A1C} - 46.7

\]

Rearranged to estimate A1C from mean glucose:

\[

\text{A1C} = \frac{\text{mean glucose} + 46.7}{28.7}

\]

8.2 Worked example

Assume a patient’s mean CGM glucose is 154 mg/dL. This is an illustrative input, not measured from a specific dataset in the retrieved sources.

\[

\text{A1C} = \frac{154 + 46.7}{28.7} = \frac{200.7}{28.7} \approx 6.99\%

\]

So a mean CGM glucose of 154 mg/dL corresponds to an estimated A1C of approximately 7.0%.

8.3 Sensitivity analysis

The derivative of estimated A1C with respect to mean glucose is:

\[

\frac{d\text{A1C}}{d\text{glucose}} = \frac{1}{28.7} \approx 0.0348

\]

Therefore, each +1 mg/dL bias in mean glucose changes estimated A1C by about +0.035 percentage points.

• A +10 mg/dL bias raises estimated A1C by about +0.35 percentage points.
• A −10 mg/dL bias lowers estimated A1C by about −0.35 percentage points.

This sensitivity is linear under the ADA formula and assumes the formula applies to the population and that mean glucose is stationary. It does not assume a specific time-in-range value. TIR and mean glucose are related but not identical; a 70% TIR target [S3] does not automatically equate to 154 mg/dL mean glucose. The calculation is therefore explicitly a mean-glucose-to-A1C calculation, not a TIR-to-A1C derivation.

8.4 Parameter table

Contradictions and source-bound claims

1. Regulatory use vs. potential consumer marketing. FDA cleared Stelo for adults 18+ not using insulin, such as those with non-insulin-treated diabetes or those without diabetes who want wellness insight [S6]. If any marketing implies use for insulin-dependent glycemic management or replacement for prescribed CGM, that would conflict with the stated FDA clearance unless a separate cleared indication is cited. This is a source-bound regulatory contradiction risk.

1. RCT efficacy vs. real-world effectiveness. Brown et al. report improved TIR in closed-loop AID [S2]. That is a controlled-trial result. DexCom’s 10-K risk-factor disclosure shows the company itself warns about adoption, reimbursement, and acceptance risks [S7]. Therefore, a statement that “trial success guarantees commercial success” would be an unsupported leap.

1. ISO 15197 category mismatch. ISO 15197:2013 governs blood-glucose self-monitoring systems [S5]. It does not govern CGM. If a claim says “CGM meets ISO 15197,” the source must be inspected to see whether the company applied the standard only to a built-in blood-glucose meter or made an analogous claim; otherwise, the claim is likely category confusion.

1. Patent vs. performance. US 6,558,351 B1 is a granted patent [S1], but a patent is a recorded claim, not proof that the covered system is superior, safe, or commercially viable. This note uses the patent only for IP-moat/feasibility context, not for performance claims.

Disconfirming Evidence

The thesis would be weakened or invalidated if any of the following were established with primary evidence:

1. Clinical benefit disappears in broad real-world populations. If independent real-world registries or subsequent randomized trials show no meaningful TIR/HbA1c advantage over non-AID care, the clinical utility claim would fail.

1. Sensor accuracy falls below acceptable performance in independent testing. A peer-reviewed or regulator-validated dataset showing unacceptable CGM accuracy in key glucose ranges, especially hypoglycemia, would undermine the safety and decision-support value of CGM/AID.

1. Reimbursement or OTC adoption collapses. If payers remove coverage, or OTC Stelo-class products fail to scale, the recurring-sensor business model would weaken, regardless of technical merit.

1. IP position proves non-defensible. If the relevant granted patents are invalidated, expire without meaningful modern continuations, or competitors demonstrate non-infringing alternatives at lower cost, the moat claim fails.

1. AID systems increase severe hypoglycemia or ketoacidosis in real-world use. The clinical value proposition depends on safety as well as TIR. Independent safety signals showing excess hospitalizations would be decisive contrary evidence.