We Searched a Wearable Glucose Monitor Idea for Prior Art — Here's What We Found
Published on patent-gnome.com/blog
Here’s the idea, exactly as it was entered, no embellishment:
A wearable biometric patch that continuously monitors human blood glucose levels through the skin using near-infrared light sensors, completely independent of needles. It transmits the real-time continuous data to a smartphone application via Bluetooth, which triggers an audible alert if the levels fall outside a pre-configured safe range.
It’s a clean, specific concept. Non-invasive glucose monitoring has been a holy grail in health tech for over a decade — no needles, no finger pricks, just a patch and a phone. The kind of idea that feels both obviously valuable and, to anyone who hasn’t checked, obviously available to build.
We ran it through Patent Gnome. It scanned over 200 patents from the EPO global database. Here’s exactly what came back.
The result: 4 direct collisions
Four patents were flagged RED — direct collision. Not “something to keep an eye on.” Direct structural overlap with the core mechanism described in the idea.
1. US2017164878A1 — Wearable Technology for Non-Invasive Glucose Monitoring
This one is about as close as it gets. The patent describes a wearable device for non-invasive glucose monitoring built around electromagnetic energy emitters and receivers. Same device category, same core sensing approach as the NIR patch in the idea. If a patent existed specifically to describe this exact concept, this would be a strong candidate.
2. WO2025166022A1 — Multimodal Physiological Monitoring System
This patent’s claim structure is worth paying attention to, because it shows exactly how claims narrow — or fail to narrow — around an idea. Claim 1 covers a wearable physiological monitoring system with optical sensors. That’s broad. Claim 2 narrows it specifically to near-infrared light. That’s the patch’s exact sensing mechanism. Claim 7 adds a flexible adhesive patch — which is the physical form factor described in the idea, not just the sensing method.
Three claims, three points of overlap, all matching the same idea. This is what a real collision looks like when you read past the abstract: it’s not one vague similarity, it’s a stack of specific claim elements that all line up.
3. WO2024173947A1 — Health Sensor Using Multiple Light Emitting Diodes
This one covers a non-invasive continuous biometric sensor using near-infrared spectroscopy, with photoemitters and photoreceivers, for determining biomolecule concentration — explicitly including glucose. Strip away the patent language and this is describing the same sensing principle as the idea: NIR light in, glucose concentration out, continuously.
4. US2018214088A1 — System and Method for Obtaining Health Data Using a Neural Network
This is the collision that’s easiest to underestimate. Claim 1 covers a device using light reflected from skin tissue — including infrared wavelengths — to determine glucose concentration. The idea adds Bluetooth transmission to a smartphone app with alerts. That sounds like a meaningful addition. It isn’t, structurally. Adding a notification layer on top of an already-claimed sensing mechanism doesn’t create a new invention — it creates a feature on top of someone else’s claimed invention. The patent’s claim doesn’t care whether you bolt on Bluetooth afterward; the core measurement step is what’s protected, and that’s what the idea’s patch does.
That last one is a pattern worth internalizing: adding software, connectivity, or UI on top of a patented physical mechanism does not get you around the patent. This is one of the most common mistakes inventors make when self-assessing freedom to operate — “but mine also has an app” is not a structural delta if the underlying sensing or mechanical claim is already covered.
The result: 2 clear
Two patents came back as contextual landscape only — meaning they’re in the same general space, but don’t structurally collide with the specific mechanism in the idea.
WO2026080527A1 predicts glucose state using heart rate and heart rate variability data — not near-infrared optical sensing. Different signal, different mechanism. A fundamentally different approach to the same problem.
WO2026079566A1 determines glucose using electrical parameters, not near-infrared light. Electrical sensing versus optical sensing is a structural delta, not just a wording difference — these are genuinely different ways of solving the same problem, and the patent’s claims are scoped to the electrical approach specifically.
These two are useful for a different reason than the four collisions: they show what doesn’t collide, and why. If the idea pivoted to using HRV-based prediction or an electrical sensing method instead of NIR optics, these wouldn’t be a problem — the structural delta that protects you from one patent is exactly what creates a new path around it.
What this means for someone with this idea
If you had this exact idea and ran a casual Google search, you might find dozens of glucose-monitoring startups and exactly zero of these four patents — because none of them are products, news articles, or company websites. They’re patent filings, sitting in patent office databases, invisible to a normal search engine, fully enforceable.
Four direct collisions on a single idea is not a death sentence, but it is a clear signal: this is a crowded, actively-patented mechanism. Before building a prototype, the inventor behind this idea would need to do one of a few things:
- Read the actual claims on all four patents (not just the summaries above) and determine if there’s a genuinely different implementation path that avoids every independent claim
- Talk to a patent attorney about freedom-to-operate, because four overlapping patents in active prosecution or grant is exactly the kind of situation where a $30 search should lead to hiring a $1,500+ professional — not replace one
- Pivot the mechanism, the way the two “clear” results suggest is possible — electrical sensing or HRV-based prediction instead of NIR optics
What this search does not tell you is whether you’d actually infringe if you built and sold this device — that’s a legal conclusion, and it requires a human attorney reading claim language against your actual implementation. What it does tell you, in under a minute, is whether that conversation with an attorney is even worth having, and exactly which patents to bring to it.
The takeaway
This is the entire value of a prior art search before you build: not a verdict, but a map. Four named patents, four specific reasons each one collides, and two examples of what a clear structural delta actually looks like. That’s enough information to make a real decision — keep going as-is, talk to a lawyer, or change the mechanism — instead of guessing.
You can see the full result, exactly as it came back, here: Wearable Glucose Monitor — Full Result →
Got an idea? Check if it’s already patented. Patent attorneys charge $1,500–$4,000 and take months. Patent Gnome takes minutes. $30 per search.