micropatch Home ←
Research brief · sources checked 14 October 2026

How a patch that reads your levels and doses your medicine would work

What Micropatch would need to measure, how it would get a drug through skin, how it would decide the dose, and how far each part has got. Every figure links to its source. Where a source is a company's own claim or a small study, the reference says so.

The short answer

An insulin Micropatch is a hard engineering problem, not an open scientific one. Glucose sensors, insulin pumps and dosing algorithms are all on sale, and in the US each has its own device class so that they can be combined.123 Nobody sells them as one patch. The closest attempt is a patch pump with a glucose sensor built into its insulin cannula. It has been tested in 18 adults.4

Wearing it for one day at a time is a design choice. Today's glucose sensors last 5 to 15 days and insulin pods last 3.567 Patches changed every 24 hours already exist for other drugs, and for insulin in the simple mechanical V-Go.8910

Dosing other drugs to their measured level in your body is still research. The first human test of a skin sensor for a drug level had six volunteers and good data for about 12 hours.11 Outside glucose, we found no approved product that measures and doses on its own.

01The problem

Between 30% and 50% of medicines for long-term illness are not taken as prescribed.12 The often-quoted WHO figure of 50%13 rests on studies from 1965 and 1975.14

Diabetes is the first target. 589 million adults aged 20 to 79 had diabetes in 2024, and over 90% of them have type 2.15 About 9.5 million people have type 1, and all of them need insulin.16 In a US registry of 22,697 people with type 1, only 21% of adults reached the HbA1c target of under 7%.17

Dosing by hand goes wrong. In the US, insulin-related lows and dosing errors caused an estimated 97,648 emergency department visits a year in 2007 to 2011, and 29.3% of those patients were admitted to hospital.18

02Sensing: what a patch can read

Glucose sensors on the skin do not read blood. They read interstitial fluid, the fluid between cells just under the skin. Glucose reaches it 5 to 6 minutes after it reaches the blood.19 That delay is small enough for a dosing algorithm to work with.

What it readsStatusEvidence
GlucoseOn saleDexcom G7 15 Day has an average error (MARD) of 8.0% and lasts 15.5 days.5 FreeStyle Libre 3 Plus lasts 15 days and is 21 mm across and 2.9 mm thick.2021
Glucose, by microneedleCleared 2025Biolinq Shine puts seven microelectrodes in the upper dermis of the forearm. It lasts 5 days, needs 2 hours to warm up and shows a colour band, not a number. It is cleared for adults who do not use insulin.6
KetonesCleared 2026Abbott's Libre Duo reads glucose and ketones in one 10-day sensor, with a US launch planned for late 2026.22
LactateHuman studiesA microneedle sensor tracked blood lactate in 21 people for 4 hours (r = 0.94).23
Sweat chemistryLabOne flexible patch read glucose, lactate, sodium, potassium and skin temperature in sweat.24 Sweat glucose is about 100 times more dilute than blood glucose and shifts with sweat rate.25
Drug levelsFirst human pilotAptamer sensors read four drugs in rats' blood every 3 seconds.26 In 2026 a microneedle patch read the antibiotic vancomycin every 5 minutes in 6 volunteers, with most usable data in the first 12 hours.11 A sensor, controller and pump held the cancer drug doxorubicin steady in rats and rabbits.27 A newer sensor chemistry lasted a week in animals.28
Heart rate, breathing, temperatureOn saleChest patches such as VitalPatch record ECG, heart rate, breathing rate and skin temperature for 5 days.29

Glucose is the only marker ready to drive a dose today. Ketones would add a safety check, because they rise when the body is short of insulin, for example when delivery has failed.

03Delivery: getting a drug through skin

Skin is built to keep things out. The outer layer is 10 to 20 µm thick. Below it, the living epidermis is 50 to 100 µm and has no blood vessels. The dermis beneath is 1 to 2 mm thick and holds the capillaries that carry a drug into the body.30

RouteStatusWhat it can do
Passive patchOn saleWorks for small, fat-soluble drugs under about 500 daltons, given in milligrams a day.3130 Rivastigmine and rotigotine patches are changed every 24 hours.89 Insulin is far too large to pass this way.
Hollow microneedlesCleared deviceNanoPass MicronJet600 injects into the skin through three 600 µm hollow needles.32
Coated or dissolving microneedlesHuman trialsA dissolving measles–rubella patch matched the injection in infants.33 Coated arrays usually carry under 1 mg of drug.34 FDA rejected Zosano's coated migraine patch in 2020 because drug exposure varied between batches.35
Hydrogel microneedlesLabThey swell with skin fluid and open a path to a separate reservoir, so the dose is not limited to what coats the needle.36
IontophoresisWithdrawnA small current pushes charged drugs through the skin. The Zecuity migraine patch was withdrawn after burns and scars.37 The IONSYS fentanyl patch was suspended in the EU over corrosion that could switch it on.38
On-body pumpOn saleOmnipod 5 holds 200 units of insulin and is worn for up to 72 hours.7 V-Go is a mechanical insulin patch worn for 24 hours.10
Glucose-responsive insulinAnimals onlyA microneedle patch that releases insulin as glucose rises controlled glucose in mice,39 and for over 20 hours in minipigs.40 Novo Nordisk's NNC2215 insulin became 3.2 times more active as glucose rose,41 but a paracetamol by-product binds it more strongly than glucose does.42

Micropatch would use a micro-pump feeding hollow microneedles or a short soft cannula. A pump can give any amount at any minute, which a coated or dissolving needle cannot. Microneedles between 480 and 1,450 µm long caused 5 to 40% of the pain of a standard 26-gauge needle. Longer needles hurt more; more needles hurt only a little more.43

04Deciding the dose

Automated insulin delivery already runs this loop. Every 5 minutes the controller reads the sensor, predicts glucose and adjusts the dose. Omnipod 5 predicts 60 minutes ahead.44 Controllers use one of three approaches: PID, which reacts to how far, how long and how fast glucose is off target; model predictive control, which forecasts glucose and plans doses; or fuzzy-logic rules.45

  • Control-IQ (iDCL trial, 168 people): time in range rose from 61% to 71%, against 59% on a standard pump.46
  • Omnipod 5 (pivotal trial): time in range rose 15.6 percentage points in children and 9.3 in adults.47
  • iLet bionic pancreas, set up from body weight alone with no carb counting: HbA1c fell from 7.9% to 7.3%, with no change on standard care.4849

The international target is more than 70% of the day between 70 and 180 mg/dL, under 4% below 70 and under 1% below 54.50

In the US these parts are regulated as connectable devices: the sensor (iCGM), the pump (ACE pump) and the controller (iAGC), all Class II.123 The controller rules require safe behaviour when communication drops and a log of every command, but they set no numeric dose limit. Each maker sets its own.3

05Fitting the dose to your body

Your insulin needs come down to a few numbers: how far one unit lowers your glucose (insulin sensitivity), how many grams of carbohydrate one unit covers (carb ratio), and how much background insulin you need each hour. Clinicians set these by hand and adjust them at appointments. A closed loop can re-estimate them from every reading. Omnipod 5 already recalculates its background dose at every pod change.44

Personal models help. In a 2025 trial of 72 people, tuning each person's settings on a computer model of their own body, a digital twin, raised time in range from 72% to 77%.51 FDA accepted the UVA/Padova simulator, with 300 virtual patients, in place of animal tests for closed-loop algorithms.52

Beyond insulin, genes matter. Of 487,409 UK Biobank participants, 99.5% carried genotypes that may cause an atypical response to at least one drug.53 Prescribing by genotype cut clinically relevant side effects from 28.6% to 21.5% in a trial of 6,944 patients.54 FDA drug labels carry gene warnings for drugs including clopidogrel, codeine and warfarin.55

Dosing by body size alone often misses. When the chemotherapy drug 5-FU was dosed by body surface area, only 20.3% of 434 patients reached the target blood level, and 60.6% were under-dosed.56 For the antibiotic vancomycin, guidelines already ask for dosing to a measured exposure target.57

06Why once a day

Today's devices are changed every 3 to 15 days. A daily patch is a deliberate trade.

For

  • Smaller. A day of insulin needs a fraction of Omnipod's 200-unit, three-day reservoir.7
  • A familiar habit. Rivastigmine and rotigotine patches are changed every 24 hours, and so is the V-Go insulin patch.8910
  • Fits drug sensors as they are. The first human drug-level sensor gave its best data in its first 12 hours.11
  • A fresh start each morning. A new sensor and dose site, and a daily update of your personal settings.

Against

  • Warm-up. Sensors need 27 minutes (G7) to 2 hours (Biolinq) before their first reading.586 Micropatch would overlap patches so the new one warms up while yesterday's still runs.
  • Skin. 28% of CGM users report skin reactions and 3.2% stop because of them.59 The rotigotine label says not to reuse a site within 14 days.9
  • Cost and waste. Ten to fifteen times as many devices as a 10 to 15 day sensor.
  • Insulin does not require it. Insulin in a pump reservoir is labelled for up to 7 days below 37°C.60

07Beyond diabetes

Outside glucose, we found no approved product that measures and doses on its own. Devices either sense or dose, and a clinician closes the loop. These are the cases with real evidence behind one half of the loop.

ConditionSenseDose
Parkinson's diseaseNo continuous sensor yet.Vyalev, a 24-hour levodopa infusion under the skin, was approved in October 2024.61 It added 1.75 hours a day of good symptom control over tablets, but 27% had redness at the site and 22% stopped because of side effects.62
Hospital antibioticsVancomycin read in skin fluid in a first human pilot.11 Penicillin tracked by microneedles in 10 volunteers.63Guidelines dose vancomycin to a measured target.57 A closed loop has been shown in rats.64
ChemotherapyNo wearable sensor for 5-FU.Only 20% of patients dosed by body size reached the target level.56
Heart failureAn implanted pressure sensor, not a patch, cut hospital admissions by about 30%.65The Furoscix on-body infusor gives 80 mg of furosemide over 5 hours.66
High blood pressureCuffless blood pressure devices are not yet recommended for diagnosis or management.67No wearable dosing.

08Regulation

  • A combination product. In the US a patch that holds both a drug and a device is a combination product. FDA picks the lead centre by the product's primary mode of action,68 and manufacturing must meet both drug and device rules.69 FDA's Office of Combination Products would decide which centre leads for Micropatch.
  • US device classes. Sensor, pump and controller are each Class II with special controls.123 Connected devices must also have a plan for security flaws, a secure design process and a software bill of materials under Section 524B.70
  • EU. Closed-loop insulin delivery and pumps with built-in sensors are Class III under Rule 22, a higher class than in the US.71 A single-use product that combines a drug and a device, such as a patch, is authorised as a medicine, with a notified body opinion on the device part.72
  • Time and fees. FDA aims to decide 90% of premarket approval applications within 180 review days.73 The fee for one in fiscal 2027 is $636,732.74 By June 2026 FDA had granted 1,320 Breakthrough Device designations, and 210 of those devices had reached the market.75

09Risks and open problems

  • Reading glucose where insulin goes in. Insulin delivered beside a sensor can disturb its reading. The one combined device tested in people reported an average error of 11.6% in 18 adults, from a company release.4
  • Sensor faults. A Class I recall of FreeStyle Libre 3 sensors that gave falsely low readings had 860 serious injuries and 7 deaths reported by January 2026.76 FDA warned Dexcom in 2025 over a changed sensor component.77 Measured against fingersticks, current sensors erred by up to 16.6%, against 11.6 to 12.0% in the lab.78
  • Blocked delivery. 3.3% of 22,741 insulin infusion sets were removed within 72 hours for prolonged high glucose.79
  • Adhesive allergy. An acrylate in early Libre sensors sensitised 12 of 15 patients tested.80
  • Hacking. In 2019 about 4,000 US patients' MiniMed pumps were recalled because a nearby attacker could change their settings, and the pumps could not be patched.81
  • Meals. Most systems still need you to tell them you are eating. iLet removed carb counting,49 but no cleared system handles meals with no input at all.
  • Privacy. Health apps and devices outside HIPAA fall under the FTC's Health Breach Notification Rule.82
  • Every new drug starts again. Each needs its own stability data at body temperature, a sensor that lasts a day in people, and its own trials.

10Who already sells the parts

CompanyWhat it sells2025 revenue
Abbott Diabetes CareFreeStyle Libre sensors$8.0B, +17.5%83
DexcomG7 sensors$4.66B, +16%84
MiniMedTubed pumps and sensors$3.10B (year to April 2026)85
InsuletOmnipod patch pumps$2.7B, +30.7%86
Tandemt:slim and Mobi pumps$1.0B, +12%87
Beta BionicsiLet bionic pancreas$100M, +54%88

One analyst puts the glucose sensor market at $13.4 billion in 2025, growing to $41.4 billion by 2033.89 That is an estimate, but it is close to Abbott and Dexcom's combined sales.

11About the simulated day

The chart on the home page is an illustrative model. It is not patient data and not a validated simulator.

  • Glucose rises with carbohydrate and falls with insulin. Insulin action peaks at 75 minutes and lasts 6 hours; meals peak at 50 minutes.
  • Each body is three numbers. Insulin sensitivity 56, 45 or 38 mg/dL per unit; carb ratio 12, 10 or 8.5 g per unit; background need 0.8, 0.9 or 1.0 units an hour, rising 30% in the early morning.
  • The walk makes insulin 70% more effective for nearly three hours.
  • Fixed doses give everyone the typical body's settings: 0.9 units an hour, plus one unit per 10 g at main meals.
  • Micropatch reads a noisy sensor running about 5 minutes behind, forecasts 75 minutes ahead allowing for insulin still working, pauses insulin when the forecast drops below 90 mg/dL and adds small corrections above 120.
  • Lows. Below 62 mg/dL the person eats 15 g of sugar.

It was tuned so the closed loop lands near the 71% time in range seen in trials, not above it.46

References

89 sources, read on 14 October 2026. Notes in amber mark company releases, small studies, animal work and other limits.

  1. 121 CFR 862.1355. Integrated continuous glucose monitoring system. Source ↗
  2. 221 CFR 880.5730. Alternate controller enabled infusion pump. Source ↗
  3. 321 CFR 862.1356. Interoperable automated glycemic controller. Source ↗
  4. 4PharmaSens / Pacific Diabetes Technologies. First-in-human study of integrated insulin delivery and glucose monitoring, April 2026. Company release; 18 adults; investigational. Source ↗
  5. 5Dexcom. Dexcom G7 15 Day receives FDA clearance, 10 April 2025. Company release. Source ↗
  6. 6US FDA. De Novo decision summary DEN240080 (Biolinq Shine), 2025. Source ↗
  7. 7Insulet. Omnipod 5 pharmacy and product information. Company page. Source ↗
  8. 8Novartis. Exelon Patch (rivastigmine transdermal system) label, FDA, 2016. Source ↗
  9. 9UCB. Neupro (rotigotine transdermal system) prescribing information. Source ↗
  10. 10V-Go wearable insulin delivery device: review. PubMed 25904143, 2015. Source ↗
  11. 11First-in-human microneedle aptamer sensor for vancomycin. PubMed 41639270, 2026. Six volunteers. Source ↗
  12. 12Kini V, Ho PM. Interventions to improve medication adherence: a review. JAMA, 2018. Source ↗
  13. 13World Health Organization. Adherence to long-term therapies, 2003. Often-quoted 50% figure traces back to studies from 1965 and 1975. Source ↗
  14. 14Mathes T et al. Citation trace of the 50% adherence figure. GMS, 2012. Source ↗
  15. 15International Diabetes Federation. IDF Diabetes Atlas, 11th edition, 2025. Source ↗
  16. 16Ogle GD et al. Global type 1 diabetes prevalence, 2025. Diabetes Res Clin Pract, 2025. Source ↗
  17. 17Foster NC et al. State of type 1 diabetes management and outcomes from the T1D Exchange in 2016–2018. Diabetes Technol Ther, 2019. Source ↗
  18. 18Geller AI et al. National estimates of insulin-related hypoglycemia and errors leading to emergency department visits. JAMA Intern Med, 2014. US data from 2007–2011. Source ↗
  19. 19Basu A et al. Time lag of glucose from intravascular to interstitial compartment in humans. Diabetes, 2013. Source ↗
  20. 20Accuracy of 15-day FreeStyle Libre 2 Plus and 3 Plus sensors. PubMed 40183340, 2025. Source ↗
  21. 21Abbott. Size of the FreeStyle Libre 3 Plus sensor. Company page. Source ↗
  22. 22Abbott. FDA authorisation for dual glucose–ketone sensing, 25 August 2026. Company release. Source ↗
  23. 23Microneedle lactate sensor in ICU, emergency and exercise settings. PubMed 41636637, 2026. Source ↗
  24. 24Gao W et al. Fully integrated wearable sensor arrays for multiplexed in situ perspiration analysis. Nature, 2016. Source ↗
  25. 25Review of the limits of sweat as a proxy for blood chemistry. PubMed 35062598. Source ↗
  26. 26Arroyo-Currás N et al. Real-time measurement of small molecules directly in awake, ambulatory animals. PNAS, 2017. Source ↗
  27. 27Mage PL et al. Closed-loop control of circulating drug levels in live animals. Nat Biomed Eng, 2017 (author blog post). Read via the authors' blog, not the paper. Source ↗
  28. 28XNA aptamer sensors extend in-body operation to one week. PubMed 41748491, 2026. Animal study. Source ↗
  29. 29VitalConnect gets FDA nod to extend VitalPatch life to five days. MobiHealthNews, 2017. Source ↗
  30. 30Prausnitz MR, Langer R. Transdermal drug delivery. Nat Biotechnol, 2008. Source ↗
  31. 31Bos JD, Meinardi MM. The 500 Dalton rule for the skin penetration of chemical compounds and drugs. Exp Dermatol, 2000. Source ↗
  32. 32NanoPass MicronJet600, FDA 510(k) K092746, 2010. Read on a third-party mirror of the FDA database. Source ↗
  33. 33Measles–rubella microneedle patch in infants, phase 1/2. Lancet, 2024. Source ↗
  34. 34Kim YC, Park JH, Prausnitz MR. Microneedles for drug and vaccine delivery. Adv Drug Deliv Rev, 2012. Source ↗
  35. 35Zosano Pharma. Form 8-K: FDA complete response letter for Qtrypta, October 2020. Source ↗
  36. 36Donnelly RF et al. Hydrogel-forming microneedle arrays for enhanced transdermal drug delivery. Adv Funct Mater / PubMed 23606824. Source ↗
  37. 37US FDA. FDA evaluating the risk of burns and scars with Zecuity migraine patch. Source ↗
  38. 38European Medicines Agency. Recommendation to suspend the marketing authorisation of Ionsys, 2008. Source ↗
  39. 39Yu J et al. Microneedle-array patches loaded with hypoxia-sensitive vesicles provide fast glucose-responsive insulin delivery. PNAS, 2015. Mice. Source ↗
  40. 40Yu J et al. Glucose-responsive insulin patch for the regulation of blood glucose in mice and minipigs. Nat Biomed Eng, 2020. Animal study. Source ↗
  41. 41Hoeg-Jensen T et al. Glucose-sensitive insulin with attenuation of hypoglycaemia (NNC2215). Nature, 2024. Animal study. Source ↗
  42. 42Paracetamol metabolite interference with a glucose-sensitive insulin. PubMed 42312167, 2026. Source ↗
  43. 43Gill HS et al. Effect of microneedle design on pain in human volunteers. Clin J Pain, 2008. Source ↗
  44. 44Insulet. How the Omnipod 5 algorithm works. Company page. Source ↗
  45. 45Review of control algorithms for automated insulin delivery. PubMed 31462875, 2019. Source ↗
  46. 46Brown SA et al. Six-month randomized, multicenter trial of closed-loop control in type 1 diabetes (iDCL). N Engl J Med, 2019. Source ↗
  47. 47Brown SA et al. Multicenter trial of a tubeless, on-body automated insulin delivery system (Omnipod 5). Diabetes Care, 2021. Single-arm trial. Source ↗
  48. 48Bionic Pancreas Research Group. Multicenter, randomized trial of a bionic pancreas in type 1 diabetes. N Engl J Med, 2022. Source ↗
  49. 49US FDA. FDA clears new insulin pump and algorithm-based software (iLet), 19 May 2023. Source ↗
  50. 50Battelino T et al. Clinical targets for CGM data interpretation: recommendations from the International Consensus on Time in Range. Diabetes Care, 2019. Source ↗
  51. 51Digital-twin-guided adaptation of automated insulin delivery settings, randomised trial. PubMed 40329052, 2025. Source ↗
  52. 52Kovatchev BP et al. In silico preclinical trials: a proof of concept in closed-loop control of type 1 diabetes. J Diabetes Sci Technol, 2009. Source ↗
  53. 53McInnes G et al. Pharmacogenetics at scale in UK Biobank. Clin Pharmacol Ther, 2020. Source ↗
  54. 54Swen JJ et al. A 12-gene pharmacogenetic panel to prevent adverse drug reactions (PREPARE). Lancet, 2023. Open-label trial. Source ↗
  55. 55US FDA. Table of pharmacogenomic biomarkers in drug labeling (updated 12 August 2026). Source ↗
  56. 56Li et al. 5-FU exposure under body-surface-area dosing. ESMO Open, 2023. Source ↗
  57. 57Rybak MJ et al. Therapeutic monitoring of vancomycin for serious MRSA infections: revised consensus guideline. Am J Health Syst Pharm, 2020. Source ↗
  58. 58Garg SK et al. Accuracy and safety of Dexcom G7 CGM in adults with diabetes. Diabetes Technol Ther, 2022. Source ↗
  59. 59Prevalence of skin reactions among CGM and insulin pump users. PubMed 36763338, 2023. Survey. Source ↗
  60. 60NovoLog (insulin aspart) label, DailyMed. Source ↗
  61. 61FDA approves AbbVie's Vyalev for adults with advanced Parkinson's disease. PharmTech, October 2024. Source ↗
  62. 62Soileau MJ et al. Safety and efficacy of continuous subcutaneous foslevodopa–foscarbidopa. Lancet Neurol, 2022. Source ↗
  63. 63Rawson TM et al. Microneedle biosensors for real-time penicillin monitoring in humans. Lancet Digit Health, 2019. Ten volunteers. Source ↗
  64. 64Closed-loop feedback-controlled vancomycin dosing in rats. ACS Sensors, 2019. Animal study. Source ↗
  65. 65Abraham WT et al. Wireless pulmonary artery haemodynamic monitoring in chronic heart failure (CHAMPION). Lancet, 2011. Source ↗
  66. 66Furoscix (furosemide injection) prescribing information, FDA. Source ↗
  67. 67Stergiou GS et al. European Society of Hypertension recommendations on cuffless blood pressure devices. J Hypertens, 2022. Source ↗
  68. 6821 CFR 3.4. Designation of agency component for combination products. Source ↗
  69. 6921 CFR 4.4. Current good manufacturing practice for combination products. Source ↗
  70. 70US FDA. Cybersecurity in medical devices: frequently asked questions (Section 524B). Source ↗
  71. 71MDCG 2021-24 Rev.1. Guidance on classification of medical devices, April 2026. Source ↗
  72. 72European Medicines Agency. Medicinal products used in combination with a medical device. Source ↗
  73. 73US FDA. MDUFA V performance goals and procedures. Source ↗
  74. 74US FDA. Medical device user fees. Source ↗
  75. 75US FDA. Breakthrough Devices Program (figures to 30 June 2026). Source ↗
  76. 76US FDA. Abbott removes certain FreeStyle Libre 3 and 3 Plus sensors (Class I recall). Source ↗
  77. 77US FDA. Warning letter to Dexcom, Inc., 4 March 2025. Source ↗
  78. 78Eichenlaub M et al. Head-to-head accuracy of three CGM systems. J Diabetes Sci Technol, 2025. Source ↗
  79. 79Kanapka LG et al. Infusion set failures in insulin pump users. Diabetes Technol Ther, 2022. Source ↗
  80. 80Herman A et al. Isobornyl acrylate as a cause of allergic contact dermatitis from FreeStyle Libre. Contact Dermatitis, 2017. Source ↗
  81. 81US FDA recall record: Medtronic MiniMed 508 and Paradigm insulin pumps, 2019. Source ↗
  82. 82US FTC. FTC finalizes changes to the Health Breach Notification Rule, April 2024. Source ↗
  83. 83Abbott. Fourth-quarter and full-year 2025 results, 22 January 2026. Source ↗
  84. 84Dexcom. Fourth-quarter and fiscal-year 2025 results. Source ↗
  85. 85MiniMed. Fiscal 2026 fourth-quarter results (Form 8-K exhibit 99.1). Source ↗
  86. 86Insulet. Fourth-quarter and full-year 2025 results, February 2026. Source ↗
  87. 87Tandem Diabetes Care. Fourth-quarter 2025 results (Form 8-K exhibit 99.1). Non-GAAP figure. Source ↗
  88. 88Beta Bionics. Fourth-quarter and full-year 2025 results. Source ↗
  89. 89Grand View Research. Continuous glucose monitoring market report. Analyst estimate. Source ↗

Cite this brief

Title
How a patch that reads your levels and doses your medicine would work
Date
14 October 2026

BibTeX

@misc{micropatch-research-2026,
  title        = {How a patch that reads your levels and doses your medicine would work},
  author       = {{Micropatch}},
  year         = {2026},
  month        = oct,
  howpublished = {Research brief},
  note         = {Sources checked 14 October 2026},
  url          = {https://micropatch.org/research.html}
}