1313811

research-article2025
DSTXXX10.1177/19322968251313811Journal of Diabetes Science and TechnologyRothenbühler et al

Original Article

Journal of Diabetes Science and Technology

A Prospective Pilot Study Demonstrating

1­–8
© 2025 Diabetes Technology Society

Noninvasive Calibration-Free Glucose Article reuse guidelines:

Measurement sagepub.com/journals-permissions
DOI: 10.1177/19322968251313811
https://doi.org/10.1177/19322968251313811
journals.sagepub.com/home/dst

Martina Rothenbühler, PhD1, Aritz Lizoain, MSc1,

Fabien Rebeaud, PhD2 , Adler Perotte, MD, MA2 ,
Marc Stoffel, PD3 , and J. Hans DeVries, MD, PhD3

Abstract
Background: Glucose is an essential molecule in energy metabolism. Dysregulated glucose metabolism, the defining feature
of diabetes, requires active monitoring and treatment to prevent significant morbidity and mortality. Current technologies
for intermittent and continuous glucose measurement are invasive. Noninvasive glucose measurement would eliminate
this barrier toward making glucose monitoring more accessible, extending the benefits from people living with diabetes to
prediabetes and the healthy.

Methods: A novel spectroscopy-based system for measuring glucose noninvasively was used in an exploratory, prospective,
single-center clinical study (NCT06272136) to develop and test a machine learning-based computational model for continuous
glucose monitoring without per-subject calibration. The study design blinded the development investigators to the validation
analyses.

Results: Twenty subjects were enrolled. Fifteen were used for the development set, and five in the validation set. All study
participants were adults with insulin-treated diabetes and median glycated hemoglobin (HbA1c) of 7.3% (interquartile range
[IQR] = 6.7-7.7). The computational model resulted in a mean absolute relative difference (MARD) of 14.5% and 96.5% of
the paired glucose data points in the A plus B zones of the Diabetes Technology Society (DTS) error grid. The correlation
between the average model sensitivity by wavelength and the spectrum of glucose was 0.45 (P < .001).

Conclusions: Our findings suggest that Raman spectroscopy coupled with advanced computational methods can enable
continuous, noninvasive glucose measurement without per-subject invasive calibration.

Keywords
glucose monitoring, noninvasive method, Raman spectroscopy

Introduction The real-time tracking of glucose via continuous glucose

monitors (CGMs) has improved the management of DM,
Glucose is a tightly regulated energy source for human cells, particularly in type 1 and insulin-treated type 2 DM, where
fueling many cellular processes and metabolic functions. reductions of glycated hemoglobin (HbA1c), increases in
Blood glucose concentration is affected by many factors, time spent in the target glucose range, and reductions in the
including food, satiety, exercise, sleep, hormones, and number of hypoglycemic episodes have been demon-
mood.1 Daily blood glucose fluctuations are a normal strated.4-7 More recent data show that benefits are also seen
response to dietary intake and energy expenditure, but an in insulin-naive type 2 DM,7,8 whereas the value of CGM in
excess caloric intake, particularly rich in sugars and starches,
combined with a sedentary lifestyle, are proven risk factors 1
Diabetes Center Berne, Bern, Switzerland
for developing dysglycemia and, eventually, diabetes melli- 2
Liom Health AG, Pfaeffikon, Switzerland
tus (DM).2 DM is a notable health concern worldwide, cur- 3
Profil, Neuss, Germany
rently affecting approximately half a billion people, projected
Corresponding Author:
to reach 783 million by 20453 with significant morbidity and J. Hans DeVries, MD, PhD, Profil, Hellersbergstr. 9, D-41460 Neuss,
mortality primarily caused by micro- and macrovascular Germany.
complications. Email: [email protected]
2 Journal of Diabetes Science and Technology 00(0)

prediabetes9 and people without dysglycemia10,11 is an area near-infrared light response signals from the epidermis and
of active research. Supporting this trend, the Food and Drug dermis, which contain the instantaneous and specific signa-
Administration (FDA) has recently cleared the use of CGMs tures of the elements composing the skin tissues and liquids.
for people not diagnosed with diabetes.12 The measured signals are sorted by wavenumbers, and then
Today’s CGMs are minimally invasive, relying on a the corresponding amplitudes are recorded before being pro-
needle-based applicator to insert a microneedle filament, cessed and analyzed offline with our computational proce-
which uses an electrochemical approach to quantify the glu- dure for glucose concentration estimation. The optics of the
cose concentration in the interstitial fluid. Since their market device were configured such that the dermis is measured,
entry, CGM devices have steadily increased their accuracy focusing primarily on measurements of interstitial fluid.
and lifetime, reduced their footprint, and expanded their con-
nectivity to phones, smartwatches, and pumps. Although
Participants
CGMs have greatly improved the quality of care for people
living with insulin-treated DM, noninvasive CGM technol- Study participants were consenting adults living with type 1
ogy can eliminate the discomfort and risks associated with or insulin-treated type 2 diabetes for at least one year and
minimally invasive monitoring, thus eliminating a signifi- meeting all the inclusion and exclusion criteria (Supplemental
cant barrier to adoption.13 However, to eliminate this barrier, Table 1). At the screening, their HbA1c was below or equal to
a noninvasive monitoring approach should also not require 9.5%. They were to have a skin tone in the Fitzpatrick scale
per-person invasive calibration to continuously measure glu- type 1 to 4,15 and a wrist exempt from injury, infection, tat-
cose accurately. Many technologies and approaches have too, or any atypical skin conditions.
been proposed and pursued, but a noninvasive method for
glucose measurement without per-person calibration has
been elusive.14
Study Procedures
We here report the first results of an exploratory, prospec- Enrolled subjects were required to fast for at least 8 hours
tive, single-center, multiple sequential-cohort study designed before the measurement procedures. An indwelling catheter
to assess the ability of a novel noninvasive investigational for blood sampling and intravenous insulin injection was
device to measure glucose without per-person calibration in inserted into one arm, and the investigational device was
adults living with type 1 or type 2 DM. fixed to the other arm. Study participants with fasting
plasma glucose above 150 mg/dL received an intravenous
insulin bolus for plasma glucose lowering. Data collection
Methods with the investigational device started 30 minutes before
This study (NCT06272136) was conducted at Profil (Neuss, the baseline plasma glucose measurement. For comparison
Germany) under consideration of the Declaration of Helsinki purposes, venous blood samples were collected using the
and in compliance with the Guideline for Good Clinical indwelling catheter during the intervention for 5 hours.
Practice and the applicable national regulations and provi- Time windows for blood sampling were every 5 minutes
sions. The study protocol was approved by the Ärztekammer from intervention until post-meal plasma glucose drop to
Nordrhein (Düsseldorf, Germany) Ethics Committee and <150 mg/dL, and from then every 15-minute interval until
exempted from approval by the German Federal Institute for the end of the intervention. The meal test period started
Drugs and Medical Devices (BfArM). Data shown here are with plasma glucose below 150 mg/dL, and the participant
from the cohorts I and II participants, recruited between received a predefined carbohydrate-rich meal without insu-
February 21 and June 4, 2024. lin administration at the start of meal intake. Plasma glu-
cose values were monitored until the level rose above 300
mg/dL. At plasma glucose above 300 mg/dL, an individual-
Study Devices ized insulin bolus dose was administered subcutaneously in
The investigational device (“Clinical Demo 2.0”, Liom the abdominal area. In case plasma glucose did not rise
Health AG, Pfaeffikon SZ, Switzerland) is a stand-alone above 300 mg/dL after meal intake, the insulin bolus was
device with a single millimeter-size exit glass window allow- administered as soon as plasma glucose stopped rising and
ing measurement on the palmar side of the subject’s wrist reached a plateau. An additional insulin bolus was adminis-
(Figure 1). The device employs Raman spectroscopy (<3500 tered in case plasma glucose did not drop below 150 mg/dL
cm−1) to capture the signature of the glucose molecule in the after insulin bolus administration. A follow-up visit was
human skin in contact with the window. During the measure- made five to eight days after the intervention day for safety
ments, the skin is continuously exposed to a sub-millimeter purposes. Adverse events, including device-related events,
near-infrared light spot at power levels below 100 milliwatts. were collected from enrollment until the end of the post-
The device collects and captures with a spectral analyzer the intervention follow-up.
Rothenbühler et al 3

Figure 1. (a) Study diagram. (b) Analysis design. (c) Investigational device with the subject’s wrist positioned on the top. A laser
interrogates the palmar side of the wrist.

Sample Size durations and glucose levels in different glycemic states. In

addition, we also report rates of absolute changes, which
As this was an exploratory clinical investigation to evaluate we categorized into “low” (less than 1.0 mg/dL/min),
the ability and the accuracy of the investigational device to “moderate” (between 1.0 and 2.0 mg/dL/min), “high”
measure glucose noninvasively and to capture preliminary (between 2.0 and 3.0 mg/dL/min), and “very high” (above
information on the investigational device to adequately plan 3.0 mg/dL/min).
further steps of clinical development, no sample size estima-
tion was performed, and no formal hypothesis testing was
planned. Computational Model Development
and Validation
Baseline and Procedural Characteristics Among the 20 subjects of this first study results, 15 were
Baseline characteristics, including demographics and infor- allocated to the training data set, although the remaining five
mation on diabetes management, are presented as medians were allocated to the validation data set. No subject was
(interquartile range—IQR) for continuous variables and withdrawn, discontinued, or lost to follow-up (Figure 1a).
counts (%) for categorical variables. Procedural character- The data used for algorithm training and development was
istics are also presented as medians (IQR) and comprise not eligible for evaluating the study objectives, and the data
4 Journal of Diabetes Science and Technology 00(0)

Table 1. Baseline Characteristics.

Overall Development cohort Validation cohort

Participants 20 15 5
Women 10 (50%) 9 (60%) 1 (20%)
Age (years) 41 (IQR = 31-45) 45 (IQR = 32-50) 31 (IQR = 27-41)
Skin color according to Fitzpatrick scale
Type I 0 (0%) 0 (0%) 0 (0%)
Type II 5 (25%) 5 (33.33%) 0 (0%)
Type III 8 (40%) 8 (53.33%) 0 (0%)
Type IV 7 (35%) 2 (13.33%) 5 (100%)
Height (cm) 173.5 (IQR = 168-178) 174 (IQR = 168-178) 173 (IQR = 169-173)
Weight (kg) 78 (IQR = 66-90) 80.5 (IQR = 61-91) 75.5 (IQR = 75-89)
BMI (kg/m2) 25.8 (IQR = 22.9-28.6) 25.3 (IQR = 22.8-28.85) 26.5 (IQR = 25.2-27.8)
HbA1c (mmol/L) 7.3 (IQR = 6.7-7.7) 7.5 (IQR = 7-7.7) 6.7 (IQR = 6.7-7.1)

Numbers shown are counts (%) or medians (IQR).

Abbreviations: BMI, body mass index; IQR, inter-quartile range.

split was predefined prior to the study start. The computa- five zones for the full validation dataset and separately per
tional procedure is a differentiable regression model which participant.
was developed based on the data of this development
cohort. The procedure processes the raw signals via simple
frequency transformation and filtering to ensure that the data In Vitro Measurement of Glucose
are standardized before being analyzed by a model for esti- A 64-mM glucose solution was prepared in a UV cuvette
mating glucose. The development investigators (FR and AP) using the same setup as in the clinical investigation to collect
were blinded to the reference data used to evaluate this a clean Raman spectral signature of the glucose molecule.
study’s primary and secondary outcomes. A predefined, The spectra were cropped in the wave number range 227 to
mutually exclusive data split was used to evaluate the pri- 3668 cm−1 and denoised.
mary and secondary endpoints of the study: the first 15 sub-
jects were assigned to the development set, and the
subsequent five to the validation set. The analysis of the vali-
Results
dation set was performed in silico, after study completion, by The study population consisted of ten women and ten men.
two independent statisticians from another group not The median age was 41 years (IQR = 31-45), the median
involved in the clinical intervention or data collection (MR body mass index (BMI) was 25.8 kg/m2 (IQR = 22.9-28.6),
and AL) (Figure 1b). All analyses were performed using the and the median HbA1c was 7.3% (IQR = 6.7-7.7). Participants
intention-to-treat data set using the statistical software R ver- had type 1 DM for more than one year and skin type II (25%),
sion 4.4.0. No imputation for missing data was performed, III (40%), or IV (35%) according to the Fitzpatrick scale15
although obviously erroneous data points based on spectral (Table 1). Among participants, the time from the start of
quality were excluded from the analyses. measurement procedures to hyperglycemia ranged from 5 to
75 minutes, and the median time in hyperglycemia was 260
Point Accuracy Assessment minutes (IQR = 197.5-274.3). Median plasma glucose val-
ues were 117 mg/dL (IQR = 104-135) at the start of the pro-
The primary analysis was performed by visualizing both the cedure and similar at discharge (119 mg/dL; IQR = 85-180).
Clinical Demo 2.0 values and the reference plasma glucose Median plasma glucose during hyperglycemia “(defined as
measurements over the time of intervention in individual glucose values above 180 mg/dL)” was 256 mg/dL (IQR =
graphs. The determination of the accuracy of the glucose 240-279). The lowest measured plasma glucose value among
estimations against venous plasma glucose measurements all interventions was 37 mg/dL, and the highest was 406 mg/
was assessed by computing both the overall mean absolute dL. Distributions across the different rate of change (RoC)
relative difference (MARD) and the MARD at the partici- categories were comparable in the two cohorts: in both
pant level. A Diabetes Technology Society (DTS) error grid groups, 16% of the measured datapoints were during very
was generated for the full validation data set for point accu- high RoC, approximately 15% in high, 31% in moderate, and
racy.16 We report the proportion of data points for each of the 37% during low RoC (Table 2). The mean absolute RoC was
Rothenbühler et al 5

Table 2. Procedural Characteristics.

Overall Development cohort Validation cohort

Subjects (n) 20 15 5
Time to hyperglycemia (min) 52.5 (IQR = 23.75-60) 55 (IQR = 47.5-60) 25 (IQR = 10-35)
Time in hyperglycemia (min) 260 (IQR = 197.5-274.25) 220 (IQR = 185-274.5) 265 (IQR = 255-270)
Time from hyperglycemia to end of 22.5 (IQR = 0-72.5) 55 (IQR = 0-82.5) 15 (IQR = 5-25)
intervention (min)
Blood glucose at study start (mg/dL) 117 (IQR = 104-135) 117 (IQR = 103-130) 132 (IQR = 106-144)
Blood glucose during hyperglycemia (mg/dL) 256 (IQR = 240-279) 256 (IQR = 240-279) 263 (IQR 247-263)
Blood glucose at discharge (mg/dL) 114 (IQR = 58-180) 104 (IQR = 58-186) 133 (IQR = 106-146)
Rate of change (mg/dL/min) 1.75 (IQR = 0.6-2.4) 1.75 (IQR = 0.60-2.40) 1.73 (IQR = 0.60-2.40)
Low RoC (counts) 426 (37.4%) 329 (37.2%) 97 (37.9%)
Moderate RoC (counts) 351 (30.8%) 272 (30.8%) 79 (30.9%)
High RoC (counts) 177 (15.5%) 139 (15.7%) 38 (14.8%)
Very high RoC (counts) 186 (16.3%) 144 (16.3%) 42 (16.4%)

Numbers shown are counts (%), medians (IQR), or minimum/maximum values.

Abbreviation: IQR, inter-quartile range.

1.73 mg/dL/min (IQR = 0.60-2.40) for the validation cohort, words, as nontarget biomarkers induce changes in the spec-
and 1.75 mg/dL/min (IQR = 0.60-2.40) for the development trum, the output of the model must be insensitive to these
cohort. directions of change. This phenomenon is also observed in
The MARD in the validation cohort was 14.5%, ranging similar approaches to model interpretation in explanation
from 11.1% to 22.0% at the individual level. The time series vectors for classification models.17 Owing to this dual role, it
of the development and validation cohorts were similar is expected that the correlation coefficient has a moderate
(Supplemental Figure 1). The individual time series of the value. We found a Pearson correlation coefficient of 0.45
five subjects of the validation cohort visually depicts the cor- (P < .001).
relation between the reference and test device glucose values In summary, 96.5% of the matched pairs between esti-
(Figure 2). The MARD in hyperglycemia (above 180 mg/dL) mated and reference plasma glucose values were in zones A
was 12.4% (range at the individual level: 7.0%-19.9%) and and B of the DTS error grid, which represent the zones where
the MARD in euglycemia (70-180 mg/dL) was 25.9% (range errors are expected to have little or no effect on clinical out-
at the individual level: 12.1%-40.8%). The MARD during come. The accuracy was relatively consistent across differ-
low RoC was 14.9%, 11.6% during moderate, 11.2% during ent RoC, ranging from an MARD of 11.2% during high and
high, and 19.6% during very high RoC. 19.6% during very high RoC.
In the validation set, 78.8% of matched pairs were in zone
A, 17.7% in zone B, and 3.5% in zone C, while none were in
Discussion
zones D and E, as per the DTS error grid16 (Figure 3). Four
of the five participants had estimated glucose values that Our study’s findings suggest that measuring glucose nonin-
were in zones A or B only, and the individual proportion of vasively without invasive calibration is possible and confirm
data points in zone A only ranged from 63.2% to 88.6% (data that optical systems combined with advanced computational
not shown). models are highly promising.
Owing to the nature of these noninvasive measurements, The strengths of this study are (1) the prospective, exter-
there is a question of whether the estimates of glucose are nal validation design where a pre-defined subset of the study
based on a correlate of glucose or based on a measurement of subjects is used to assess the performances of the computa-
glucose itself. To investigate this, we evaluated our model’s tional model issued from a development cohort, (2) the focus
average glucose estimate sensitivity to changes in Raman on developing a calibration-free computational model, and
spectra (ie, the gradient of the independent variable of glu- (3) the study procedure that consisted of rapid and large glu-
cose with respect to the dependent variables of Raman spec- cose excursions in a population of people living with pro-
trum intensities) and compared this to the glucose spectrum found defects in their glucose metabolism. There is a rich
as evaluated from a series of aqueous glucose solutions literature of approaches and attempts to measure glucose
(Figure 4). To provide robust estimates of glucose, the model noninvasively, and the typical approach is to develop a
must not only be sensitive to glucose but also insensitive to computational model with per-person calibration. This fact
other Raman signals unrelated to glucose estimation. In other is often implicit in the design of the development and
6 Journal of Diabetes Science and Technology 00(0)

Figure 2. Time series of the five subjects of the validation cohort.

Individual time series of the five subjects of the validation set. In black, the reference glucose values, and in pink, the glucose values from the
investigational device.

validation measurements. In this case, the development and glucose values. This model, applied to all time periods for
validation measurements were partitioned by subject such the remaining subjects, accurately tracked the full trajectory
that no subjects were present in both the development and despite never being exposed to decreasing glucose values
validation cohorts. In addition, the evaluation of the valida- (data not shown).
tion cohort estimates was done in an independent fashion Overall, the performance is comparable to first-genera-
such that strictly separate investigator groups performed tion CGMs18 but unlike those systems, this device does not
model development and model evaluation. The study popu- require per-person calibration. Today’s best-in-class CGMs
lation has large and rapid glucose changes that make glucose are factory-calibrated and do not require users to perform
monitoring especially challenging. invasive measurements, such as fingerstick blood glucose
Limitations include the modest sample size of the valida- measurements, unless CGM results do not match user symp-
tion cohort, the lack of subjects with darker skin (types V and toms. Their success, combined with their limitation of inva-
VI on the Fitzpatrick scale), the lower glucose peak than siveness, has long spurred interest in developing noninvasive
expected, and the limited data in hypoglycemia. As the glu- methods to sense glucose continuously. The glucose mole-
cose profiles used for training and validation had a very simi- cule’s well-characterized optical properties have led to mul-
lar shape, it could be that the model learned to output the tiple attempts to develop spectroscopic sensors. Although
shape of the profile, independent from the input. Therefore, numerous proofs-of-concept have been shown in laboratory
we developed a similar model with data from only a subset of conditions, this is one of the first to demonstrate state-of-the-
the subjects and only for time periods of strictly increasing art performance without invasive calibration.19 In addition,
Rothenbühler et al 7

to have the potential for a convenient, comfortable wearable

form factor, such technology should have the potential for
miniaturization while still demonstrating adequate accuracy
and robustness in real-life conditions.14,20,21 The convergence
of recent developments in the fields of photonics, electron-
ics, and computational sciences, especially regarding minia-
turization, opens new opportunities to bring continuous,
noninvasive glucose sensing to the wrist eventually. The
results presented here lay down the foundations toward the
development of such a technology.

Conclusions
This study shows that the Raman spectroscopy-based tech-
nology embedded in the investigational device and the asso-
ciated machine-learning computational model bear great
promise to enable the development of a wearable, noninva-
sive glucose monitor that does not require per-person cali-
bration. Future efforts will focus on miniaturizing the
technology to a smartwatch-like form factor to ensure user
convenience and further developing and testing the computa-
tional model to ensure adequate accuracy is also achieved in
Figure 3. (a) Diabetes Technology Society (DTS) error grid.
(b) Number and percent of paired data points in the different the real-life utilization of a CGM, in a more diverse popula-
zones of the DTS error grid. tion, including the assessment of the system’s performance
Abbreviations: DTS, Diabetes Technology Society; EG, error grid. in healthy and prediabetes populations.

Figure 4. Model sensitivity compared with glucose spectrum.

Abbreviations Acknowledgments
CGM, continuous glucose monitor; DM, diabetes mellitus; DTS, The authors are grateful for all the work and continuous support
Diabetes Technology Society; IQR, interquartile range; BfArM, from many colleagues at Liom Health AG, Profil GmbH, and DCB
German Federal Institute for Drugs and Medical Devices; MARD, Research AG, without which this work would not have been
mean absolute relative difference; RoC, rate of change; HbA1c, gly- possible.
cated hemoglobin.
8 Journal of Diabetes Science and Technology 00(0)

Declaration of Conflicting Interests 8. Aleppo G, Hirsch IB, Parkin CG, et al. Coverage for continuous
The author(s) declared the following potential conflicts of interest glucose monitoring for individuals with type 2 diabetes treated
with respect to the research, authorship, and/or publication of this with nonintensive therapies: an evidence-based approach to
article: AP and FR are the employees of Liom Health AG. policymaking. Diabetes Technol Ther . 2023;25(10):741-751.
9. Cichosz SL, Kronborg T, Laugesen E, et al. From stability to
Funding variability: classification of healthy individuals, prediabetes,
and type 2 diabetes using glycemic variability indices from
The author(s) disclosed receipt of the following financial support continuous glucose monitoring data. Diabetes Technol Ther.
for the research, authorship, and/or publication of this article: This 2024;27:33-44. doi:10.1089/dia.2024.0226.
work was funded by Liom Health AG. 10. Freckmann G, Hagenlocher S, Baumstark A, et al. Continuous
glucose profiles in healthy subjects under everyday life con-
ORCID iDs ditions and after different meals. J Diabetes Sci Technol.
Fabien Rebeaud https://orcid.org/0009-0007-6336-3158 2007;1(5):695-703.
11. Holzer R, Bloch W, Brinkmann C. Continuous glucose moni-
Adler Perotte https://orcid.org/0000-0002-6695-0282
toring in healthy adults: possible applications in health care,
Marc Stoffel https://orcid.org/0000-0003-4048-2318 wellness, and sports. Sensors. 2022;22:2030.
J. Hans DeVries https://orcid.org/0000-0001-9196-9906 12. Commissioner of the FDA. FDA clears first over-the-counter
continuous glucose monitor. FDA. 2024. Accessed March 5,
Supplemental Material 2024. https://www.fda.gov/news-events/press-announcements/
fda-clears-first-over-coun ter-continuous-glucose-monitor.
Supplemental material for this article is available online.
13. Engler R, Routh TL, Lucisano JY. Adoption barriers for con-
tinuous glucose monitoring and their potential reduction with
References a fully implanted system: results from patient preference sur-
1. Hermanns N, Ehrmann D, Shapira A, Kulzer B, Schmitt A, veys. Clin Diabetes. 2018;36(1):50-58.
Laffel L. Coordination of glucose monitoring, self-care behav- 14. Ahmed I, Jiang N, Shao X, et al. Recent advances in opti-
iour and mental health: achieving precision monitoring in dia- cal sensors for continuous glucose monitoring. Sens. Diagn.
betes. Diabetologia. 2022;65(11):1883-1894. 2022;1:1098-1125.
2. Mozaffarian D, Kamineni A, Carnethon M, Djoussé L, 15. Fitzpatrick TB. The validity and practicality of sun-reactive
Mukamal KJ, Siscovick D. Lifestyle risk factors and new-onset skin types I through VI. Arch Dermatol. 1988;124(6):869-
diabetes mellitus in older adults: the cardiovascular health 871.
study. Arch Intern Med. 2009;169:798-807. 16. Klonoff DC, Freckmann G, Pleus S, et al. The Diabetes
3. Sun H, Saeedi P, Karuranga S, et al. IDF Diabetes Atlas: Technology Society error grid and trend accuracy matrix for
global, regional and country-level diabetes prevalence esti- glucose monitors. J Diabetes Sci Technol. 2024;18(6):1346-
mates for 2021 and projections for 2045. Diabetes Res Clin 1361.
Pract. 2022;183:109119. 17. Baehrens D, Schroeter T, Harmeling S, Kawanabe M, Hansen
4. Danne T, Nimri R, Battelino T, et al. International consen- K. How to explain individual classification decisions. J Mach
sus on use of continuous glucose monitoring. Diabetes Care. Learn Res. 2010;11:1803-1831.
2017;40:1631-1640. 18. Cappon G, Vettoretti M, Sparacino G, Facchinetti A.
5. Peters AL, Ahmann AJ, Battelino T, et al. Diabetes technology: Continuous glucose monitoring sensors for diabetes manage-
continuous subcutaneous insulin infusion therapy and continu- ment: a review of technologies and applications. Diabetes
ous glucose monitoring in adults: an Endocrine Society clinical Metab J. 2019;43(4):383-397.
practice guideline. J Clin Endocrinol Metab. 2016;101:3922- 19. Hirsch IB, Tirosh A, Navon A. Noninvasive real-time glu-
3937. cose monitoring is in the near future. Diabetes Technol Ther.
6. Edelman SV, Argento NB, Pettus J, Hirsch IB. Clinical impli- 2024;26(9):661-666.
cations of real-time and intermittently scanned continuous glu- 20. Laha S, Rajput A, Laha SS, Jadhav R. A concise and system-
cose monitoring. Diabetes Care. 2018;41(11):2265-2274. atic review on non-invasive glucose monitoring for potential
7. Jancev M, Vissers TACM, Visseren FLJ, et al. Continuous diabetes management. Biosensors. 2022;12:965.
glucose monitoring in adults with type 2 diabetes: a system- 21. Alsunaidi B, Althobaiti M, Tamal M, Albaker W, Al-Naib I.
atic review and meta-analysis. Diabetologia. 2024;67(5):798- A review of non-invasive optical systems for continuous blood
810. glucose monitoring. Sensors. 2021;21:6820.