Non-invasive glucose monitoring: which startup is ahead? Last updated: 31 July 2026 In our wearable technology market deck , you will find everything you need to understand the market SUMMARY Photonic View Technology is ahead in non-invasive glucose monitoring today, but its lead is narrow and still rests on research hardware rather than a finished everyday product. No company leads every part of the race. Photonic View has the strongest large calibration-free study, RSP Systems has the deepest repeated-use evidence, and Liom has the most credible wearable strategy. Raman spectroscopy has quietly taken control of the scientific leaderboard. The top three companies all use it, which suggests the field is moving toward direct molecular measurement even though the optics remain difficult to shrink and manufacture. The lowest headline error does not belong to the strongest study. HAGAR reported a 6.7% MARD, but Photonic View’s larger cohort, laboratory reference and calibration-free setup make its 14.6% result more convincing for a mass-market device. Regulatory approval has not settled the technology question. Jingce Medical and Bobang Fangzhou already sell approved products in China, yet both remain supplementary tools with calibration or accuracy limitations that prevent them from replacing fingersticks or conventional CGMs. Liom may still become the consumer winner. It has the most disclosed funding, the clearest plan for continuous wrist-worn monitoring and a 2028 launch target, but its published independent validation remains much smaller than the company’s product ambition. Calibration is the dividing line between an impressive prototype and a scalable product. A device that needs repeated finger-prick setup can still be useful, but it gives up much of the convenience that makes non-invasive monitoring attractive in the first place. The companies best prepared to manufacture today are not the ones with the strongest underlying science. Jingce and Bobang have regulated products and supply chains, while Photonic View, Liom and HAGAR are still trying to turn more promising sensing methods into reliable hardware. The biggest competitive threat may come from outside the startup field. Apple, Samsung, Abbott and Dexcom have much larger engineering, regulatory and manufacturing resources, so proprietary human datasets and hard-to-reproduce hardware knowledge will matter more than patents alone. Our conclusion is straightforward: Photonic View leads the science and the overall ranking, RSP is closest on evidence, and Liom has the best shot at building the product people actually want. None has yet proved calibration-free, multi-day continuous monitoring with the reliability of a conventional CGM. This market map, featured in our wearable technology market deck , highlights top companies and startups in the wearable technology market Non-invasive glucose monitoring: which startup is actually ahead? Which non-invasive glucose-monitoring startups are really in the race? The credible non-invasive glucose-monitoring field currently has nine startups with enough human data, regulatory progress or working hardware to deserve a serious comparison. We include Photonic View Technology, RSP Systems, Liom, HAGAR, Jingce Medical, BOYDSense, DiaMonTech, Bobang Fangzhou and Afon Technology. They use methods ranging from Raman and infrared spectroscopy to radiofrequency waves, metabolic heat, microwaves and breath analysis, but all are trying to estimate glucose without puncturing the skin. We use a strict definition. Biolinq, Nemaura and other microneedle companies are excluded because their sensors still enter the skin. Apple, Samsung and Huawei are also outside the startup comparison, and none currently sells an authorized watch that directly measures glucose without skin penetration. The FDA continues to warn consumers that it has not authorized any smartwatch or smart ring making that claim. Occuity and Synex Medical remain worth watching, although neither has released enough comparable human glucose data to rank confidently. Know Labs is also left out because its ownership and strategy changed enough to make the former standalone glucose business difficult to assess. Lepu Medical has relevant products, but it is an established public medical-device company rather than a startup. Funding is unusually hard to compare in this category. Some companies report grants, some disclose only one round, and several Chinese startups never publish a cumulative total. The table therefore shows the minimum amount we could verify publicly. Startup What it is building Where it stands now Publicly disclosed funding Photonic View Technology Calibration-free Raman measurement using depth-selective micro-SORS Large clinical study completed; first medical product in development At least RMB 200 million, plus later undisclosed rounds RSP Systems Raman spectroscopy through the skin Portable prototypes and repeated clinical studies About 25 million Jingce Medical Metabolic-heat device measured beneath the tongue NMPA-approved and marketed in China Not publicly disclosed BOYDSense Breath-based glucose estimation Clinical study completed; portable breath prototype €7 million disclosed capital increase DiaMonTech Mid-infrared photothermal glucose measurement CE-certified desktop platform; portable D-Pocket still in development More than $20 million, plus later undisclosed financing Bobang Fangzhou Metabolic-heat glucose meter measured at the finger NMPA-approved and sold in China At least RMB 50 million from one disclosed round; total unknown Afon Technology Microwave wrist wearable called Glucowear Early real-world prototype testing More than £2 million in public grant support Is any non-invasive glucose-monitoring startup clearly ahead today? Photonic View is narrowly ahead overall today, with RSP Systems and Liom close enough that one strong trial or product milestone could change the order. The three leaders are ahead for different reasons. Photonic View has the strongest large calibration-free study. RSP has spent longer testing repeated measurements and use outside the clinic. Liom has the clearest plan for turning spectroscopy into a wearable people might actually use every day. Jingce Medical and Bobang Fangzhou lead on availability because they already have products on the Chinese market, while DiaMonTech has a longer regulatory history than most rivals. HAGAR sits just behind the top three because its radiofrequency technology may be easier to shrink, but its public clinical evidence remains limited. So the field has a scientific top three, a few companies with genuine regulatory progress, and several higher-risk challengers. Photonic View leads, but not by much. If you want more recent data on this point, please see our latest wearable technology market report . As this chart shows, and as featured in our wearable technology market deck , search interest in smart rings has been increasing rapidly Has any non-invasive glucose startup put a real product on sale? Jingce Medical and Bobang Fangzhou have sold regulated non-invasive glucose devices in China, but no startup currently offers a broadly approved replacement for fingersticks or conventional CGMs. Jingce’s combined glucometer received approval from China’s NMPA as a supplementary tool for adults with type 2 diabetes. The device places a metabolic-heat probe beneath the tongue and also contains an invasive measurement module. Public clinical disclosures report a 10.69% MARD, but a new user reportedly needs around ten invasive reference readings to build a personal calibration model. Bobang’s earlier device measures metabolic and physiological signals at the fingertip. Its multicenter study reported a 27.02% MARD, and performance weakened during parts of the post-meal period. The approved product can provide extra readings without another finger prick, but patients cannot use the non-invasive value alone to replace standard testing or adjust treatment. DiaMonTech’s desktop DMT Base obtained CE certification in 2019, yet broad commercialization never followed. The company has spent years developing the smaller D-Pocket instead. Regulatory clearance for a large clinical device did not automatically produce a practical consumer product. We have not found any startup disclosing a large active-user base, recurring glucose-monitoring revenue or widespread physician adoption. The first commercial products are real, but none has become a primary glucose monitor. Which startup has the strongest non-invasive glucose data? Photonic View has the strongest overall clinical evidence, while RSP Systems has the best repeated calibrated result and HAGAR has the lowest headline error. Photonic View’s Nature Metabolism study tested its depth-selective Raman system in a main cohort of 230 people. It reported a 14.6% MARD without personalized calibration, and 99.4% of estimates landed in the clinically acceptable zones of the consensus error grid. The study used venous plasma glucose as the reference. The device still needed about eight minutes for each reading and was roughly desktop-computer sized. Accuracy also weakened in the low-glucose range. Photonic View has shown that its method can work across a relatively broad group of people, but not yet that it works quickly or continuously in a wearable. RSP’s Scientific Reports study followed 50 adults with type 2 diabetes and produced 1,918 paired readings. Its portable Raman device reached a 12.8% MARD after ten reference measurements were used to personalize a pretrained model. RSP has also studied home use lasting up to 25 days. RSP has shown more clearly that one person can use the system repeatedly. Photonic View has shown more clearly that one model can work across many people without a long personal setup. We give Photonic View the edge because a mass-market device cannot begin with hours of finger-prick calibration. Liom’s published pilot involved 20 insulin-treated adults, with only five held out for validation. The reported MARD in that validation group was 14.5%. The result is promising, but five people cannot show how the algorithm will behave across a broad population. HAGAR reported a 6.7% MARD across 75 participants, with every reading in Clarke error-grid zones A or B. Most participants contributed very few readings, and the reference was an average of consumer fingerstick values rather than a laboratory analyzer. The result deserves attention, but it still needs tougher testing. Startup Human evidence Reported performance How much weight we give it HAGAR 75 participants, generally one or two readings each 6.7% MARD; 100% in Clarke zones A and B Striking result, but the comparator and limited repeated use make it less robust RSP Systems 50 participants and 1,918 paired readings 12.8% MARD; 100% in consensus zones A and B Strong repeated result, with a meaningful calibration burden Liom 20 participants, including five in the validation group 14.5% MARD; 96.5% in DTS zones A and B Promising and calibration-free, but based on a very small holdout sample Photonic View Main cohort of 230 participants 14.6% MARD; 99.4% in consensus zones A and B Strongest large calibration-free evidence, although still from a large, slow device Jingce Medical Multicenter clinical program disclosed around its approval 10.69% MARD; 94.72% in Clarke zone A Regulated result, but the personal invasive calibration limits comparability DiaMonTech Recent 36-person clinical evaluation Below 20% MARD Direct optical measurement, but accuracy remains behind the leading Raman systems Bobang Fangzhou 200 participants and 589 paired readings 27.02% MARD; 93.9% in consensus zones A and B Commercially real, but technically far behind the leading studies BOYDSense 130 adults, including a 30-person validation group 98.7% in Parkes zones A and B Good feasibility evidence, although no directly comparable MARD was published If you want more recent data on this point, please see our latest wearable technology market report . This chart, included in our wearable technology market deck , illustrates yearly VC funding for wearable technology startups Which startup is closest to a wearable non-invasive glucose monitor? Liom is currently closest to a credible wearable, although its best clinical data still came from a benchtop system rather than the final watch-sized device. Liom has built its strategy around continuous metabolic monitoring. The company says it has made a major miniaturization step, improved how much usable light the device captures and designed the platform around day-long battery use. It has also set a commercial launch target around 2028. The hard part begins when the hardware leaves the bench. A wearable must handle sweat, movement, changing pressure, temperature shifts and rapid glucose swings. Liom’s study already showed less consistent performance during fast changes and in parts of the normal glucose range. HAGAR may have the simpler hardware path. Its radiofrequency chip is designed for a wrist device and should fit more naturally into standard electronics manufacturing. It still needs a large dataset showing stable numerical readings across days and different users. Afon recently tested Glucowear during normal activity and reported about 75% accuracy when separating normal from elevated glucose. One wearer and a classification task remain far from a medical CGM trial, but the update shows functioning wearable hardware outside the laboratory. Photonic View is taking a slower route through a clinical device, while RSP’s current hardware and calibration process remain better suited to planned measurements. Liom leads the wearable race. HAGAR has the strongest alternative architecture. Can any non-invasive glucose startup monitor continuously without calibration? No startup has proved that a calibration-free non-invasive device can continuously track glucose with the multi-day reliability of a conventional CGM. Personal calibration remains one of the field’s hardest problems. Skin thickness, hydration, temperature, tissue composition, placement and pressure can all change the signal. A model may work after learning one person’s body and then struggle with a new user, or with the same user under different conditions. Photonic View’s published model worked without collecting each user’s glucose history first. Liom also reported calibration-free predictions, although its independent sample was small. HAGAR describes GWave as calibration-free, but its study did not follow people long enough to show whether drift or repositioning eventually forces a reset. RSP starts with ten reference readings collected over a short period. That is an improvement over the weeks of calibration used in earlier work, but it still asks a new customer to prick a finger several times. Jingce uses a similar personal-model approach, while Bobang requires an invasive reference value during setup. Continuous use adds another layer of difficulty. The sensor must stay stable through sleep, meals, exercise, sweat and motion while also catching low glucose and rapid changes. RSP has the strongest repeated-use history, including home studies lasting up to 25 days, but the system still resembles repeated scanning more than an autonomous CGM. Liom is designing for continuous operation from the start, although it has not yet published multi-day traces from the miniaturized wearable. HAGAR and Afon both describe continuous wrist monitoring, but their public human evidence remains too thin. Photonic View is still too slow in its current form, while BOYDSense, Jingce and Bobang are better understood as spot-measurement systems. Photonic View leads on calibration-free evidence, RSP leads on repeated use and Liom leads on continuous product design. Nobody has combined all three. This chart, included in our wearable technology market deck , shows why Whoop is leading in wearable technology Which sensing technology looks strongest for non-invasive glucose monitoring? Raman spectroscopy currently has the strongest human evidence, while radiofrequency sensing offers the cleaner route to a small, cheap wearable if HAGAR can validate it properly. Photonic View, RSP and Liom all use Raman spectroscopy, which looks for molecular vibrations linked to glucose. These systems try to isolate a glucose-related optical signature rather than estimate glucose mainly from heat, circulation or broader tissue changes. The approach has produced the strongest peer-reviewed human studies in our comparison. The weakness is hardware. Raman signals through skin are faint, so the device needs careful optics, enough laser power, sensitive detectors and algorithms that remove background noise. That explains why the best results still come from relatively large instruments and controlled positioning. DiaMonTech uses mid-infrared photothermal detection, another direct optical approach. The company has credible clinical work and a long regulatory history, although its recent performance below 20% MARD still trails the top Raman systems. HAGAR and Afon use radiofrequency or microwave signals. Their sensors could be smaller and easier to manufacture, but the measured signal can also change with hydration, blood flow, temperature and other tissue properties. Jingce and Bobang estimate glucose from metabolic heat and other physiological inputs. Their approvals show that indirect models can become useful medical devices, but the calibration requirements and weaker published performance expose their limits. BOYDSense is even more indirect because breath compounds can shift with food, medication, oral health and wider metabolism. If you want more recent data on this point, please see our latest wearable technology market report . Which non-invasive glucose startup is closest to broader regulatory approval? Jingce Medical and Bobang Fangzhou lead on approvals already obtained, while Photonic View looks best placed among the newer systems aiming for much better accuracy. China’s NMPA approved Jingce’s combined glucometer as a supplementary device for adults with type 2 diabetes. Bobang also obtained Chinese approval for its fingertip metabolic-heat product. Both companies completed product development, clinical work, manufacturing documentation and quality controls. Their indications remain narrow. The devices support existing glucose monitoring rather than replace it, and patients cannot rely on the non-invasive reading alone for treatment decisions. DiaMonTech’s earlier CE certification proves that an optical non-invasive system can clear a European regulatory milestone. The absence of broad commercialization afterward suggests that the first product format was not practical enough. Photonic View has the strongest package for the next serious submission: a large study, a laboratory-grade reference and no personal calibration. Its first target is a medical device rather than a watch, which reduces the initial burden around battery life, motion and continuous use. It still needs testing with the final hardware, better low-glucose performance and faster measurements. RSP brings more real-life experience but must justify its calibration process. Liom needs a much larger trial with the wearable itself. HAGAR needs to reproduce its result against a laboratory reference over several days. This chart, included in our wearable technology market deck , illustrates yearly funding for wearable technology startups Which non-invasive glucose startup has found real customer demand? No non-invasive glucose startup has yet proved strong, repeatable demand through transparent revenue, active-user or reorde
Non-invasive glucose monitoring: which startup is ahead?
NewMarketPitch Team