Clinical Results

A new optical method for blood glucose self-control


The clinical trials were conducted at the Diabetes Institute in St Joseph’s Hospital in Heidelberg, Germany and in the university ophthalmological clinic, Heidelberg, Germany. EyeSense sensors were implanted into the eyes of type 1 and type 2 diabetics under local anesthetic. The patients collected data via a small portable fluorescence photometer (fig.1). The data were recorded over several measurement sittings over a timeframe of more than six months. With each sitting, the patients were subjected to glucose stimulation (orally administered carbohydrate and delayed administration of insulin). The recorded calibrated fluorescence intensity was compared with laboratory capillary blood glucose value, measured via standard procedures.

The study was carried out over several cohorts with increasingly better sensor designKlinische Ergebnisse

Fig.1 Patient measuring with portable fluorescence photometer

Results: Sensor wearability

The minimal invasive procedure of implantation under local anesthetic was pain free and simple to conduct. The implants were tolerated well throughout the entire study with a minimum wearing time of 6 months and some patients wearing the implants for over 2 years









Fig.2 Patient with the hydrogel sensor (pale blue) in the right eye (over 6 months implantation). The same patient looking straight ahead showing the sensor is not visible (right)

Results: Glucose monitoring

After only a short wearing time (up to a month), all patients recorded extremely accurate measurements. Only a short delay time was observed between the EyeSense values and the reference values. This shows that the glucose concentration in the interstitial fluid is only minimally time delayed behind the capillary blood glucose concentration.

After a longer wearing period, a capsule formed around the implant. This led to an increase in the time delay between glucose signals and the capillary blood glucose values. In the latest trials we were however able to reduce this by applying a biocompatible coating onto the sensor. In the first week the EyeSense predicted glucose values were remarkably accurate with the mean absolute relative deviation (MARD*) being on average 7 percent (as compared with laboratory capillary blood values). After 6 months this deviation had increased, but with the sensors coated with biocompatible film only to 14 percent. The norm for in vitro blood glucose measuring devices ISO15197*, requires a deviation of less than 20 percent for 95 percent of all values.

MARD is the abbreviation of mean absolute relative deviation. For every piece of data in a measuring period, the amount of the relative deviation is recorded. (Under certain circumstances negatives are removed) and the average is calculated.

*ISO15197. In vitro diagnostic test systems: requirements for blood-glucose monitoring systems for self-testing in managing diabetes mellitus. International Organization for Standardization, 2003.


The clinical results mentioned emphasize the potential of this new technology to replace finger pricking as a method of glucose measurement. Once the sensor is implanted, this new method offers the patient a comfortable, non-invasive blood glucose measuring technique, which over a period of six months, maintains an impressive accuracy and precision.

If you are interested in further details, we look forward to you contacting EyeSense via this link.


The detailed study results have been presented at several international congresses and published in journals

  1. Müller et al., New Optical Method for Blood Glucose Self-Monitoring, Nov 2011, DTM, Los Angeles (Poster).
  2. Hasslacher et al., Safety and accuracy of a new longterm subconjunctival glucose sensor, J. Diabetes 2012; 4: 291-296.
  3. Müller et al., First Clinical Evaluation of a New Long-term Subconjunctival Glucose Sensor, J. Diabetes Sci. Technol., 2012; 6(4): 875-883.
  4. Müller et. al., Blood Glucose Self-Monitoring with a Long-Term Subconjunctival Glucose Sensor, J Diabetes Sci. Technol., 2013; 7(1): 24-34.