www.nature.

com/scientificreports

OPEN Internet of Things enabled open

source assisted real‑time blood
glucose monitoring framework
Abubeker K. M 1*, Ramani. R 2, Raja Krishnamoorthy 3, Sreenivasulu Gogula 4, Baskar. S 5,
Sathish Muthu 6, Girinivasan Chellamuthu 7 & Kamalraj Subramaniam 8

Regular monitoring of blood glucose levels is essential for the management of diabetes and the
development of appropriate treatment protocols. The conventional blood glucose (BG) testing have
an intrusive technique to prick the finger and it can be uncomfortable when it is a regular practice.
Intrusive procedures, such as fingerstick testing has negatively influencing patient adherence.
Diabetic patients now have an exceptional improvement in their quality of life with the development
of cutting-edge sensors and healthcare technologies. intensive care unit (ICU) and pregnant women
also have facing challenges including hyperglycemia and hypoglycemia. The worldwide diabetic rate
has incited to develop a wearable and accurate non-invasive blood glucose monitoring system. This
research developed an Internet of Things (IoT) - enabled wearable blood glucose monitoring (iGM)
system to transform diabetes care and enhance the quality of life. The TTGOT-ESP32 IoT platform with
a red and near-infrared (R-NIR) spectral range for blood glucose measurement has integrated into this
wearable device. The primary objective of this gadget is to provide optimal comfort for the patients
while delivering a smooth monitoring experience. The iGM gadget is 98.82 % accuracy when used after
10 hours of fasting and 98.04 % accuracy after 2 hours of breakfast. The primary objective points of the
research were continuous monitoring, decreased risk of infection, and improved quality of life. This
research contributes to the evolving field of IoT-based healthcare solutions by streaming real-time
glucose values on AWS IoT Core to empower individuals with diabetes to manage their conditions
effectively. The iGM Framework has a promising future with the potential to transform diabetes
management and healthcare delivery.

Diabetes, a chronic metabolic disorder characterized by higher blood glucose levels, demands consistent moni-
toring and intervention. According to the World Health Organization (WHO), diabetes was the largest cause
of mortality in 2021, accounting for 24.6 % of diabetic ­fatalities1. Maintaining appropriate Blood Glucose (BG)
levels is of utmost importance in mitigating the potential risks associated with heart f­ ailure2,3, kidney d ­ iseases4,
and retinal ­diseases5. Traditional glucose monitoring methods have often been cumbersome, inconvenient, and
disconnected from patients’ daily lives. Therefore, the timely detection of diabetes is of utmost importance to
mitigate, monitor, or delay the progression of the condition and its related problems. Covid’19 is more likely to
cause significant problems in people with d­ iabetes6,7. Another disorder that affects blood glucose levels is hypogly-
cemia. Individuals diagnosed with diabetes experience a compromised capacity to manage blood glucose levels,
leading to fluctuations characterized by hyperglycemia or h ­ ypoglycemia8,9. Non-invasive techniques obviate the
necessity for uncomfortable finger punctures, rendering glucose monitoring more convenient and less obtrusive
than traditional glucose monitoring. Utilizing a non-invasive approach presents a promising opportunity for

1
Department of Electronics and Communication Engineering, Amal Jyothi College of Engineering (Autonomous),
Koovappally, Kerala, India. 2Department of Computer Science and Engineering, P.S.R Engineering College,
Sivakasi, Tamilnadu, India. 3Department of Electronics and Communication Engineering, Kalasalingam Academy
of Research and Education, Krishnankoil, Tamilnadu, India. 4Department of Computer Science and Engineering,
(Data Science), Vardhaman College of Engineering, Shamshabad, Hyderabad, India. 5Faculty of Engineering,
Department of Electronics and Communication Engineering, Karpagam Academy of Higher Education,
Coimbatore, Tamilnadu, India. 6Department of Orthopaedics, Government Medical College & Hospital, Karur,
Tamilnadu, India. 7Department of Orthopaedics, Saveetha Medical College and Hospital, Saveetha Institute
of Medical and Technical Sciences (SIMATS), Saveetha University, Chennai, Tamilnadu, India. 8Department
of Biomedical Engineering, Karpagam Academy of Higher Education, Coimbatore, Tamilnadu, India. *email:
kmabubeker82@gmail.com

Scientific Reports | (2024) 14:6151 | https://doi.org/10.1038/s41598-024-56677-z 1

Vol.:(0123456789)
 www.nature.com/scientificreports/

continuously monitoring glucose levels, eliminating the necessity for regular blood collection. Using real-time
data can offer an enhanced understanding of glucose variation, facilitating prompt modifications in medication,
diet, and lifestyle.
According to recent research, individuals with diabetes mellitus (DM) are more likely to acquire pneumo-
nia, and their death rate is significantly greater than that of non-diabetic p­ atients10,11. Pre-existing diabetes has
linked to variations in blood glucose levels in individuals with community-acquired pneumonia (CAP) and a
high mortality r­ ate12. A continuous glucose monitor (CGM) is an alternative for patients with ­hypoglycemia13,14.
Non-invasive glucose monitors are in high demand since many patients are averse to repeatedly puncturing their
fingertips to check their blood glucose levels. In addition, there are some issues with the precision of current
CGM technology, so a non-invasive sensor that is more precise than current CGMs would be advantageous.
Recently, machine learning (ML), and artificial intelligence (AI) techniques has showcased the capabilities
estimating blood glucose levels. These techniques utilize data obtained from non-invasive Wearable Devices
(WD), which in turn enables the monitoring and management of individuals with diabetes using digital
­biomarkers15,16. The Vivovitals platform, developed by Sachmechi et al.17, serves as an additional tool to aid in
the reduction of HbA1c levels in individuals diagnosed with uncontrolled type 2 diabetes mellitus. This platform
is utilized in combination with lifestyle modifications and regular adjustments to antidiabetic drugs. Wear-
ability and clinical accuracy are two critical components of this kind of device’s design. Due to the near-field
sensor’s length, Kseniya Zavyalova et al.18 developed a device based on a flexible substrate. The sensor’s extended
near field and high electromagnetic field penetration depth were shown with a slot antenna combination. The
technological advancement highlighted the potential of human biological fluids, including t­ ears19, ­saliva20, and
interstitial ­fluids21, as biomarkers for monitoring glucose levels. In a study by K.A. Saraswathi et al.22, a unique
non-enzymatic electrochemical sweat sensor was developed for glucose monitoring using polyaniline nanocaps
as its foundation. Sina Kiani et al.23, developed a microwave resonant sensor operated at a frequency of 9 GHz
and incorporated a band-pass filter within a substrate-integrated waveguide cavity with split ring resonator. In
contemporary times, a nascent cohort is emerging, but mainly as a theoretical concept rather than a practical
implementation of medical devices, owing to its developing phase of advancement. However, for the sake of
this discussion, we will classify these devices as fourth-generation medical devices. This category encompasses
noninvasive techniques that provide real-time and continuous monitoring.
Figure 1 depicts the iGM- AWS IoT architecture used in this research. The proposed iGM sensor array has
been developed and deployed in ESP32 Module with Organic LED (OLED) display, open-source software with
embedded C structure, and Amazon Web Services (AWS) IoT core infrastructure. Recent advancements in sen-
sor technology, wearable devices, and data analytics have enhanced the accuracy of non-invasive blood glucose
monitoring. Non-invasive blood glucose monitoring will enhance quality of life, diabetic control, and reduction
of infection. Furthermore, it can enhance intellectual capacity of diabetes and improve therapeutic interventions.
The developed iGM gadget reduces risk factors for hyper and hypoglycemic conditions. This framework enables
patient’s lifestyles and healthcare by effectively using non-invasive sensor devices and open-source software. The
important research highlights are discussed below.

Figure 1.  Proposed AWS assisted iGM framework for non-invasive blood glucose monitoring.

Scientific Reports | (2024) 14:6151 | https://doi.org/10.1038/s41598-024-56677-z 2

Vol:.(1234567890)
www.nature.com/scientificreports/

1. Designed and deployed an IoT enabled wearable non-invasive real-time blood glucose monitoring using red
and near-infrared (R-NIR) spectroscopy.
2. The developed red and near-infrared (R-NIR) sensor framework is integrated into wearable devices for real-
time monitoring and data is streamed to Amazon web services (AWS) IoT core architecture.

The rest of this research paper is organized as follows: Sect. “Methodology” investigated the methodology
and deployment of the iGM device. Section “Experimental results” presents a detailed analysis and experimental
evaluation conducted on the iGM framework. Section “Conclusion” includes the research conclusion and the
potential areas for future research in this field.

Methodology
The authors Abubeker K ­M1, *, Dr. R. ­Ramani2, Sreenivasulu ­Gogula3, Raja ­Krishnamoorthy4, S ­Baskar5, Dr.
Sathish ­Muthu6, Dr. Girinivasan C
­ hellamuthu7, and Kamalraj ­Subramaniam8 hereby confirm that,

1. We confirm that all methods discussed in this manuscript were carried out in accordance with relevant
guidelines and regulations.
2. We confirm that all experimental protocols were done by the non-invasive methods (No finger prick or
direct blood samples are not taken for experiment) and hence no approval from institutional and/or licens-
ing committee is required.
3. We confirming that informed consent was obtained from all subjects.

The graphical abstract in Fig. 1 illustrates the open-source-assisted, non-invasive wearable IoT architecture
for the iGM framework. The proposed system designed using an ESP32 IoT platform with an inbuilt Organic
LED display, open-source software for deploying Red and near-infrared sensors with an integrated C structure,
and AWS IoT core. Each component of the iGM gadget is described in-depth in the next section.

Non‑invasive sensor deployment

This research uses red and near IR signals to improve blood glucose monitoring with IoT architecture. Low-
Level Light Therapy (LLLT) using R-NIR light is now widely used to treat diabetes problems at the cellular level.
According to clinical trials, red light treatment uses 630 nm to 660 nm 810 nm to 850 nm wavelength for NIR
signals, it has the broadest range of benefits and no adverse effects. Organic molecules are best detected using
near-infrared signals because their bonds absorb wavelengths in the NIR range. Organic chemicals are the main
constituents of biological tissue, and NIR can penetrate more profoundly than any other infrared wavelength.
Figure 2 represents the skin’s reflectance characteristic; red signal absorption is high when glucose molecules
are present. NIR light can penetrate further into the body than visible light and reach the subcutaneous arteries.
It is also possible to do multivariate analysis on the reflected light by collecting it at various optical wavelengths
at varying concentrations of blood glucose. To overcome the limitations of earlier approaches, a combined R-NIR
biosensor has developed. The suggested system is reliable for continuous practical deployment, and the sensor
location on the subject’s wrist or earlobe is significant.
Figure 2 illustrates that a decrease in glucose concentration leads to increased scattering, resulting in reduced
absorption. Conversely, increased glucose concentration in tissues leads to decreased scattering, therefore increas-
ing tissue absorption. The intensity of reflected light is lower in tissue with higher glucose content due to the
increased light absorption by high glucose tissue, as shown by the following equation.

IRi = IIi epl . (1)

According to Eq. (1), the reflected light intensity from skin tissues ­(IRi) equals the product of the reflected
light intensity ­IIi and the NIR signal’s transit span through the tissue(l). Photon absorption in tissue ­(Pa) is often
calculated as follows:

Figure 2.  Red and near-infrared (R-NIR) signal absorption characteristics of a human skin.

Scientific Reports | (2024) 14:6151 | https://doi.org/10.1038/s41598-024-56677-z 3

Vol.:(0123456789)
 www.nature.com/scientificreports/

Pa = (2.303MC), (2)
Equation (2) shows that the tissue concentration and molar extension significantly impact how many photons
the tissues absorb. Beer-Lambert law states that light intensity through a medium decrease exponentially. Fol-
lowing that, the absorption of NIR spectra is ­(Anir) depicted as follows:
Anir = log [IIi /IRi ]. (3)
The concentration of red (­ Rbg) and NIR (­ NIRbg) ranges is determined as:
Cred = log [Rbg /IIi ], (4)

Cnir = log [NIRbg /IIi ]. (5)

Following the measurement of blood glucose, the red (­ Rag) and NIR (­ NIRag) concentration of is calculated
as follows:
Cred = log [Rag /IIi ], (6)

Cnir = log [NIRag /IIi ]. (7)

The spectral concentrations in the red and NIR regions shift as seen in the following Eqs. (8) and (9).
�red = log [Rag /Rbg ], (8)

�nir = log [NIRag /NIRbg ], (9)

To minimize calibration time, a full-day blood glucose concentration change estimate was evaluated. LED
and photodiode are placed next to one other on the same tissue site for a diffused reflectance measurement.
A sensor has placed on the wrist skin with a 3 mm spacing with device body for better accuracy and reflec-
tion without affecting the external noises and disturbances. Interconnected systems of superficial veins make up
the wrist. Effective blood sampling is guaranteed by positioning the sensor in this area. By keeping the sensor
in constant touch with the skin, the wrist reduces the impact of movement artefacts on the sensor’s accuracy. In
Fig. 3, a sensor array includes a NIR LED (880 nm), a trans-impedance amplifier and a photodiode (800–1700
nm). Absorption and dispersion reduce light’s intensity when it comes into contact with living tissue. The wrist
reflects an attenuated and distorted light that is picked up by the receiver. Light noise is reduced by noise filter-
ing components, and then the received signal is amplified so that it can travel farther despite its low strength.

Figure 3.  Circuit schematic of the developed iGM gadget.

Scientific Reports | (2024) 14:6151 | https://doi.org/10.1038/s41598-024-56677-z 4

Vol:.(1234567890)
www.nature.com/scientificreports/

ESP32’s 12-bit analogue to digital converter (ADC) converts the desired signal into an electric current, and a
trans-impedance amplifier (TIA) using the Op-Amp converts the photodiode current to voltage. TIAs have a
significant propensity to oscillate at high frequencies over the gain-bandwidth product because of their high
gain operation. A 0.8Hz High Pass Filter (HPF) and a 10Hz Low Pass Filter (LPF) are used to eliminate the DC
component from the resulting R-NIR signal and to reduce the power line and other signal disturbances that
might interact with the PPG and give us inaccurate data, respectively. The iGM framework provided an IoT-based
system by merging the AWS IoT core with a personalized sensor circuit despite the rarity of the methodologies
for IoT-based systems.
Various non-invasive blood glucose measuring devices are compared in Table 1. There are a wide variety of
photoplethysmographs in use, some of which contain artificial intelligence and machine learning algorithms
to predict blood glucose. Reflection and absorption spectroscopy with fingertip measurements is the most flex-
ible technique to apply NIR spectroscopy. Researchers have yet to develop an accurate and flexible device, and
the vast majority of them are still working on PPG-based systems. The IGM framework provided an IoT-based
system by merging the AWS IoT core with a bespoke sensor circuit despite the rarity of the methodologies for
IoT based systems. Compared to other devices on the market, it boasts a high degree of precision and a low rate
of inaccuracy.

Open source library development

The future of open source development will be driven by the need for the Internet of Healthcare Things (IoHT).
There has been a dramatic increase in the use of Tactile Internet (TI) technology in recent years to speed up
real-time data delivery from many applications, including healthcare.
Figure 4 illustrates the open-source structure of the proposed iGM framework developed with embedded
C, licensed under Massachusetts Institute of Technology (MIT). Additional functionality has been added to the
sketch or source code using C++ libraries. iGM_Max.cpp has the programming library, iGM_Max.h comprises
the headers and one text file that lists all of the keywords used in library development. Custom functions in the

SI. no. Methodology Mode of operation Type of gadget Type of architecture

1 Proposed iGM gadget using IoT Real time IoT based Wearable and wrist based. AWS IoT Core
Noninvasive blood glucose monitor via multi-sensor
2 Automatic Wearable and fingertip based Hybrid and multi-layer embedded sensors
­fusion24.
Blood glucose and blood pressure from a PPG by
3 Artificial intelligence based Non wearable and fingertip based A multilayer CNN, SVM and random forests.
means of machine learning ­techniques25.
4 PPG based machine learning ­approach26. Automatic Wearable and fingertip based Embedded module and custom sensors
Near-infrared spectroscopy based on PLSR combines
5 Manual Non wearable and fingertip based Hybrid model of PLSR and SAE with DNN
SAE deep neural network ­approach27
Fingertip capillary dynamic near infrared spectrum
6 (DNIRS) measurement combined with multivariate Automatic Portable and fingertip based Dynamic NIR spectrum method
linear modification algorithm 28.
Wavelength-modulated differential photo thermal
7 Automatic Non wearable Embedded sensors
­radiometry29.
30
8 Visible-NIR optical ­biosensor . Automatic Wearable band Embedded sensors

Table 1.  Performance evaluation of different technologies used in non-invasive blood glucose measurement.

Figure 4.  Structure of iGM open source library licensed by MIT.

Scientific Reports | (2024) 14:6151 | https://doi.org/10.1038/s41598-024-56677-z 5

Vol.:(0123456789)
 www.nature.com/scientificreports/

iGM_Max.cpp files include RValue () and NIRValue () functions, which are used to return the red and near-
infrared reflection values from the inner wrist and earlobes of the patient.

AWS IoT core

A message queuing and telemetry transport (MQTT) is a machine-to-machine (M2M) communication infra-
structure and a Secure hypertext transfer protocol (HTTP) are two IoT services of AWS IoT Core for publishing
and subscribing to messages. Smart wearable sensors may send and receive data with limited bandwidth, making
the ability to exchange IoT data effectively. The iGM framework has integrated with a TTGO ESP32 IoT develop-
ment board powered by an Xtensa dual-core 32-bit microprocessor with a 240 MHz operating frequency, with
organic LED display. Wi-Fi and Bluetooth are incorporated into the ESP32, making it a suitable Iot platform
for server less connections to AWS IoT Core devices. AWS offers various services, including the MQTT library,
and AWS IoT core communication establishment. The first step is to construct the AWS IoT Things, IoT things
should record physical actions; IoT Policy is a JSON document specifying authorized operations. AWS IoT core
receives sensor data through the MQTT protocol, and the open-source library analyses it.

Algorithm 1.

ESP32 has traditionally been the best choice for developing IoT based wearable applications because of its
flexibility in open source programming, IoT connectivity, and low-power design with an OLED display. The
manuscript investigates the deployment of an open-source library for iGM based on the Arduino IDE and
includes a custom module that has been specifically defined for the derived R-NIR sensors.

Experimental results
The blood glucose level is a crucial indicator of diabetes control and is expressed as milligrams per deciliter (mg/
dL). Diabetic patients can maintain a fasting blood glucose level of 80 to130 mg/dL and a postprandial blood
glucose level of less than 180 mg/dL. A blood test called glycosylated hemoglobin (HbA1c) can estimate how
stable a person’s blood glucose has over the previous two to three months. Patients with diabetes should aim
for an HbA1c level of 48 mmol/mol or less (6.5%). The normal range for HbA1c in people with type 2 diabetes
is between 5.7 and 6.4 %. This research outcome shows the average blood glucose (ABG) levels in people with
diabetes have strongly correlate with their HbA1c values. Several variables determine the HbA1c level of an
individual such as, age, general health, pregnancy status, and diabetes background.
Table 2 shows the results of seven tests used to evaluate the effectiveness of the proposed hardware. The
subjects are selected with equal priority for men and women with different age groups including non-diabetic,
prediabetes and diabetic patients. During evaluation phase, the efficacy of the iGM device has tested within a
day with a one-hour interval, and researchers consider the various levels of participants by considering age,
diabetes history, HbA1c, etc. In contrast to CGM, the proposed iGM makes use of near-infrared technology
to detect blood glucose without intrusive procedures. It can be employed in the earlobes, finger tips, wrist, and
even soft tissues of the body, without contact with the patient’s body. The observed glucose level is shown on the
wearable device’s OLED screen, and it can additionally be broadcast to AWS IoT Core cloud enabling remote
monitoring. Table 2 contrasts the test results of 7 individuals picked at random from various groups, such as
those pre-diabetic, non-diabetic, and type 1 or type 2 diabetes. During a 10-hour fast, the iGM device’s accuracy
is 98.82 %; two hours after breakfast, it’s 98.04 %. Test participants are restricted to 15, and the error rate reach
up to 1.18 % and 1.96 %, respectively, when compared to current best practices. Equation (10) presented the
formula for calculating the error percentage (Er).
Er = [(IGMBG − RefRG ) /RefBG] × 100. (10)
Here, iGMBG and RefBG are the BG value from iGM and Accu-Check Active respectively.

Scientific Reports | (2024) 14:6151 | https://doi.org/10.1038/s41598-024-56677-z 6

Vol:.(1234567890)
www.nature.com/scientificreports/

BG level –mg/dl
Diabetic history (in Pre-prandial (before Post prandial (1.5 Hr.
Subjects Age & gender year) Diabetic status Fasting (8 AM) 2 Hr. after breakfast meals 1 PM) after meals 2.30 PM) HbA1c
S1 39-M 2 Prediabetes 128 164 134 172 5.8
S2 41-F 0 Nil 110 127 119 143 5.2
S3 53-M 14 Diabetic 169 231 200 256 6.9
S4 65-F 4.5 Diabetic 221 248 234 256 11.8
S5 48-F 6 Diabetic 183 221 205 244 7.3
S6 43-M 1.5 Prediabetes 115 153 129 168 5.4
S7 51-F 7 Diabetic 193 230 189 245 7.6

Table 2.  Estimated blood glucose levels throughout the day using iGM with pre-diabetic subject (S1,S6), non-
diabetic subject (S2) and diabetic subject (S3,S4,S5,S7).

Accuracy (Ac) of the IGM device is determined by first determining the error rate (Er) using Eq. (10). The
percentage error rate is calculated by dividing the difference between the iGM and blood glucose values by the
Laboratory testing, as shown in Eq. (11).
Ac = 100% − Er. (11)
By comparing the readings with those from an Accu-Check Active blood glucose monitor, the researchers
were able to confirm that the proposed device have superior clinical accuracy. A total of three different groups
of subjects participated in the tests; healthy volunteers between the ages of 30 and 40 without a history of blood
glucose; diabetes patients between the ages of 35 and 55, including men and women. Accu- Check Active takes
invasive blood samples from the subject’s finger while iGM collects real-time blood glucose data from patient’s
wrist and earlobe. Blood glucose levels is measured in the morning before breakfast, lunch, and dinner, and two
hours after each meal and snack. The real time BG values are monitored on IoT enabled wrist watch with OLED
display as well as on the AWS cloud platform. The ESP32 microcontroller is a powerful platform for developing
wearable applications due to its open-source development environment, IoT connectivity, and low-power design
with OLED display. The manuscript investigates the deployment of an open-source library for iGM based on the
Arduino IDE and includes a personalized module that has specifically defined for the developed R- NIR sensors.
This research used different testing with various subject matter and compare with current techniques to
empirically support the claim that the suggested approach enhances glucose monitoring accuracy using non-
invasive sensors. Independent validation, statistical significance of the iGM device, reference measurements,
etc., have all been performed in this research.
Table 3 compares the performance of the developed iGM hardware with the prominent research in this area.
When evaluating CGM devices, it is essential to consider individual preferences and needs, personal lifestyle,
daily activities, and specific medical requirements. The CGM device’s readings align with reference blood glucose
measurements from traditional finger stick tests or lab-based methods. It is observed that the iGM device does
not require frequent calibration. Researchers have compared the accuracy of the suggested approach to that of
many other laboratory tests and non-invasive blood glucose measurement techniques. Compare the upfront
and ongoing costs of the iGM system; the iGM market is continually evolving, be sure to refer to the most cur-
rent and authoritative sources for the latest information on device performance comparisons. By comparing the
readings with those from an Accu-Check Active blood glucose monitor, the researchers were able to confirm
that the suggested device had superior clinical accuracy.
Blood glucose levels in the iGM before one hour of breakfast to two hours after supper (08 am to 10 pm) for
diabetes, healthy individuals, are shown in Table 4. The developed iGM blood glucose monitor has evaluated
using one-hour intervals for analyzing the accuracy and performance of the monitor. One-hour intervals may
effectively observe fluctuations in blood glucose levels, ensuring a temporal resolution, which may help identify
patterns and fluctuations in glucose compared with less frequent monitoring. For diabetic patients, blood glucose

Mechanism Targets Advantages Disadvantages Costs Versatility

Real time data can be
IoT based monitoring, bat- Wearable and wrist or ear-
iGM Non-invasive NIR based. Type 1 & 2 diabetes. updated only with internet 30 $
tery life up to 6 months. lobe based BG monitoring.
facility.
Measure interstitial glucose Separate sensor and reader. Attached to the body with
CGM- abbott diabetes care
Invasive-interstitial type. Type 1 & 2 diabetes. and stores 90 days of glucose Expensive and 10-15 days’ 306 $ separate display with WiFi
Ltd, ­UK31
data. battery life. connectivity.
Sensor backup for up to One touch activation and
One care meta ­sense32 Invasive-interstitial type. Type 1 & 2 diabetes. Self-activate the sensor. 66 $
14 days. health coach through app
Relaying glucose data to
Glucose-based meal score
Ultrahuman ­M133 Invasive-interstitial type. Type 1 & 2 diabetes. Connects to aerobic activity. the app requires NFC con- 303 $
and optimization tips.
nection.

Table 3.  Performance comparison of the iGM and other CGM devices available in the global market.

Scientific Reports | (2024) 14:6151 | https://doi.org/10.1038/s41598-024-56677-z 7

Vol.:(0123456789)
 www.nature.com/scientificreports/

Time → (24 Hr.) 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

S1-(iGM) 128 142 164 158 139 134 159 172 143 145 132 130 141 130 130
S1-ACC​ 129 141 164 157 140 133 159 170 141 144 133 130 142 129 130
S1-LAB 128 140 165 158 1490 133 159 171 142 145 132 132 140 131 130
S2-(iGM) 110 121 127 129 126 119 131 143 120 129 120 119 129 122 119
S2-ACC​ 110 120 125 127 125 120 130 144 119 130 121 120 130 `120 120
S2-LAB 111 121 126 129 125 120 131 142 120 128 121 120 129 123 121
S3-(iGM) 169 201 231 226 209 200 239 256 221 232 218 214 232 211 204
S3-ACC​ 170 200 230 227 210 200 240 258 225 230 220 217 233 214 203
S3-LAB 170 200 230 225 210 200 240 255 220 233 220 214 232 210 204

Table 4.  Measured blood glucose levels using iGM, Accu-Check Active (ACC) device and laboratory tests
(LAB) results for subjects S1 to S3.

levels can have fast fluctuations, mostly after meals. This allows ability to detect postprandial glucose spikes
and determine the delay between fluctuations in blood glucose levels and the monitor’s recorded values. This
approach is suitable for statistical analysis, facilitating the evaluation of the monitor’s accuracy and precision. To
obtain real-time readings, the Accu-Check Active (ACC) device and laboratory testing are used in this research.
This comparison evaluated the accuracy and reliability of blood glucose measurements obtained from the iGM
system, ACC device, and laboratory tests (LAB) results.

Accuracy and precision

• iGM: The iGM system provides continuous glucose monitoring, eliminating the need for frequent finger
pricking. It measures interstitial fluid glucose levels and has shown promising accuracy compared to labora-
tory tests. From Table 3, the proposed system has high accuracy and reliability in BG measurement.
• Accu-Check: The ACC glucometer measures capillary blood glucose levels and is widely used for self-mon-
itoring by individuals with diabetes. It was extensively validated and proven accurate and reliable in many
studies. Still, as per the results, it needs more accuracy in chemical test results in the laboratory.
• LAB: Laboratory tests involving venous blood samples are considered the gold standard for blood glucose
measurement. They offer high accuracy and precision, but the process involves more time and resources than
the other two methods.

Convenience and ease of use

• iGM: The portable and self-wearable nature of the iGM system offers unparalleled convenience, as it continu-
ously monitors glucose levels without the need for frequent manual testing. However, its initial placement
does not require a medical procedure and professional calibration.
• ACC Device: The ACC glucometer is portable, user-friendly, and requires a small amount of blood for testing.
It is ideal for daily self-monitoring and provides instant results.
• LAB: Laboratory tests require individuals to visit a healthcare facility for blood sample collection, which may
not be as convenient as using the other two methods for daily monitoring.

Device cost

• The iGM system’s initial cost, including the deployment procedure and device, can be relatively less than
traditional glucometers. From Table 2 we can see that the cost of the product is only ten percentage of the
devices available in the market.
• ACC Device: The ACC glucometer is less affordable in low-income countries regarding initial cost and ongo-
ing expenses for test strips.
• LAB: Laboratory tests are typically covered by health insurance, but they can incur higher costs when per-
formed frequently.

Continuous monitoring

• The iGM system monitors glucose levels and provides a continuous glucose profile, offering valuable insights
into glucose trends over time.
• ACC Device: The ACC glucometer provides instant blood glucose readings, but it does not offer continuous
monitoring.

Scientific Reports | (2024) 14:6151 | https://doi.org/10.1038/s41598-024-56677-z 8

Vol:.(1234567890)
www.nature.com/scientificreports/

• LAB Results: Laboratory tests provide a single snapshot of blood glucose levels at the time of the test, limiting
their ability to capture fluctuations throughout the day.

The blood glucose monitoring method choice depends on individual preferences, needs, and the level of
diabetes management required. The iGM system offers the advantage of continuous monitoring but comes with
lower initial costs and no processing cost required. Ultimately, consultation with healthcare professionals can
help individuals select the most suitable method for their diabetes management plan.
Figure 5 shows an iGM device developed with R-NIR spectroscopy is deployed in patient’s wrist with OLED
display. A small data-collection window of 20 seconds was chosen to guarantee a steady but continuous system
further. Inter-integrated circuit (I2C) is an in-built protocol used for short-distance communication in embed-
ded system components, and it is used to connect the iGM model’s NIR and RLED reflection brightness to the
ESP32 Iot platform. Some earlier techniques developed to gather signals from a person’s fingers or arms have
proven helpful in blood glucose measurement systems. Fingers can be used for invasive and non-invasive blood
glucose measures; in the former scenario, a drop of blood has been extracted from a capillary. On the other
hand, several of the alternatives that have been presented are invasive and limit the subject’s mobility. So that the
proposed sensor system can function as an IoHT-compliant wearable device using non-invasive NIR technology.
Figure 6 depicts the performance comparison between the iGM and Accu Check Active blood glucose moni-
tor. The developed iGM would function as portable device for constantly measuring blood glucose levels of
ambulatory, ICU, and pregnant women without calibration. This Wearable, non-invasive glucose monitors have
improved diabetic management and enhance the lives of millions of diabetic patients worldwide. The iGM can
be used to make prompt modifications to insulin dosages and lifestyle choices with artificial intelligence and
machine learning algorithms. In addition to this better diabetes management is possible to analyze individual

Figure 5.  An IoT-enabled real-time blood glucose monitoring system (iGM) deployed in diabetic patient using
ESP32 IoT module with organic LED display.

Figure 6.  iGM vs Accu Check Active blood glucose test results comparison for non- diabetic, pre-diabetic and
diabetic patients. The iGM glucose measurement on the wrist and Accu check active on patients’ fingertip.

Scientific Reports | (2024) 14:6151 | https://doi.org/10.1038/s41598-024-56677-z 9

Vol.:(0123456789)
 www.nature.com/scientificreports/

glucose patterns and forecast future glucose trends. The use of non-invasive glucose monitoring equipment
can be extended to health and fitness apps to track the changes in food and exercise affect in blood sugar levels.

Conclusion
The proposed iGM framework helps for diabetic patients to keep continuous monitoring blood glucose levels
for improved treatment and enhanced life styles. The measured blood glucose levels can be monitored remotely
and treated promptly using information sent to physicians via IoT devices. The wearable iGM gadget uses a
customized sensor with Red Near InfraRed (R-NIR) spectroscopy and an I2C protocol interface. Open-source
frameworks improve the accuracy, user adaptability and creativity among healthcare professionals, facilitat-
ing the framework’s enhancement and modification. The research findings of the iGM blood glucose monitor
and the invasive state-of–art methods are compared with three groups of subjects throughout the research by
considering the factors like age, diabetes status, and HbA1c. The signal quality and streaming capabilities of the
iGM gadget best suited for ICU patients, and isolated patients. The main challenge of this product is the battery
backup, it can take up to 23 days, and may improve by using Li-ion batteries. Future developments will integrate
Nano sensors and deploying artificial intelligence and machine learning algorithms will enhance the gadget
performance. Also data visualization tools will heighten the user experience in interpreting glucose data and
facilitate proactive decision-making.

Data availability
The datasets used and/or analysed during the current study available from the corresponding author on reason-
able request.

Received: 15 December 2023; Accepted: 9 March 2024

References
1. WHO. https://​www.​who.​int/​news-​room/​fact-​sheets/​detail/​diabe​tes (Accessed 15 August 2022) (2021).
2. Wu, L. et al. Increased arterial stiffness elevates the risk of heart failure in diabetic patients. Int. J. Cardiol. 385, 26–33. https://​doi.​
org/​10.​1016/j.​ijcard.​2023.​05.​039 (2023).
3. Zhang, X. et al. WATCH-DM risk score predicts the prognosis of diabetic phenotype patients with heart failure and preserved
ejection fraction. c 385, 34–40. https://​doi.​org/​10.​1016/j.​ijcard.​2023.​05.​045 (2023).
4. Nabrdalik, K. et al. Diabetic peripheral neuropathy is associated with diabetic kidney disease and cardiovascular disease: The
silesia diabetes-heart project. Curr. Probl. Cardiol 48(8), 101726. https://​doi.​org/​10.​1016/j.​cpcar​diol.​2023.​101726 (2023).
5. Ouyang, S. et al. Cataract, glaucoma, and diabetic retinopathy are independent risk factors affecting falls in the older adult with
eye diseases. Geriatr. Nurs. 53, 170–174. https://​doi.​org/​10.​1016/j.​gerin​urse.​2023.​07.​001 (2023).
6. Lyons, C. E., Zhu, I. & Gill, M. K. Accelerated progression of diabetic retinopathy following severe COVID-19 infection. Am. J.
Ophthalmol. Case Rep. 101911, 2023. https://​doi.​org/​10.​1016/j.​ajoc.​2023.​101911 (2023).
7. Piarulli, F. et al. Association of renin-angiotensin-aldosterone system inhibitors with best COVID-19 outcomes in a diabetic
population of the Veneto region (north-east Italy): A lesson for endemic phase? Nutrition. Metab. Cardiovasc. Dis. https://​doi.​
org/​10.​1016/j.​numecd.​2023.​06.​016 (2023).
8. Davies, M. J. et al. Management of hyperglycemia in type 2 diabetes, a consensus report by the American Diabetes Association
(ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care 45(11), 2753–2786. https://d ​ oi.o​ rg/1​ 0.2​ 337/​
dci22-​0034 (2022).
9. Salimi, H. R., Jalili, S., Griffiths, M. D. & Alimoradi, Z. Fear of hypoglycemia and its predictive factors among diabetic pregnant
women. Primary Care Diabetes 17(1), 68–72. https://​doi.​org/​10.​1016/j.​pcd.​2022.​11.​009 (2023).
10. Hua, J., Huang, P., Liao, H., Lai, X. & Zheng, X. Prevalence and clinical significance of occult pulmonary infection in elderly patients
with type 2 diabetes mellitus. BioMed. Res. Int. https://​doi.​org/​10.​1155/​2021/​31873​88 (2021).
11. Yang, A. et al. Time-varying risk associations of renin angiotensin system inhibitors with pneumonia and related deaths in a cohort
of 252,616 patients with diabetes (2002–2019). Diabetes Res. Clin. Pract. 185, 109233. https://d ​ oi.o
​ rg/1​ 0.1​ 016/j.d
​ iabre​ s.2​ 022.1​ 09233
(2022).
12. Jensen, A. V. et al. The impact of blood glucose on community-acquired pneumonia: a retrospective cohort study. ERJ Open Res.
https://​doi.​org/​10.​1183/​23120​541.​00114-​2016 (2017).
13. Islam, M. S., Qaraqe, M. K., Belhaouari, S. & Petrovski, G. Long term HbA1c prediction using multi-stage CGM data analysis.
IEEE Sens. J. 21(13), 15237–15247. https://​doi.​org/​10.​1109/​JSEN.​2021.​30739​74 (2021).
14. Grundman, J. B. et al. Applying the use of shared medical appointments (SMAs) to improve continuous glucose monitor (CGM)
use, glycemic control, and quality of life in marginalized youth with type 1 diabetes: Study protocol for a pilot prospective cohort
study. Contemp. Clin. Trials Commun. 32, 101067. https://​doi.​org/​10.​1016/j.​conctc.​2023.​101067 (2023).
15. Ahmed, A., Aziz, S., Qidwai, U., Abd-Alrazaq, A. & Sheikh, J. Performance of artificial intelligence models in estimating blood
glucose level among diabetic patients using non-invasive wearable device data. Comput. Methods Progr. Biomed. Update 3, 100094.
https://​doi.​org/​10.​1016/j.​cmpbup.​2023.​100094 (2022).
16. Gade, A., Vijaya Baskar, V. & Panneerselvam, J. Hybrid model with optimal features for non-invasive blood glucose monitoring
from breath biomarkers. Biomed. Signal Process. Control https://​doi.​org/​10.​1016/j.​bspc.​2023.​105036 (2023).
17. Sachmechi, I. et al. Frequent monitoring of blood glucose levels via a remote patient monitoring system helps improve glycemic
control. Endocr. Pract. 29(6), 441–447. https://​doi.​org/​10.​1016/j.​eprac.​2023.​03.​270 (2023).
18. Zavyalova, K., Gorst, A. & Mironchev, A. A non-invasive determination of blood glucose concentration using a near-field sensor.
Proceedings 60, 1–5. https://​doi.​org/​10.​3390/​IECB2​020-​07022 (2020).
19. Fiore, L. et al. Paper card-like electrochemical platform as a smart point-of-care device for reagent-free glucose measurement in
tears. Chem. Commun. 59(29), 4300–4303. https://​doi.​org/​10.​1039/​d2cc0​6561d (2023).
20. Feng, L., Lin, X., Feng, J., Min, X. & Ni, Y. NiNP/Cu-MOF-C/GCE for the the noninvasive detection of glucose in natural saliva
samples. Microchem. J. 190(108657), 2023. https://​doi.​org/​10.​1016/j.​microc.​2023.​108657 (2023).
21. Tianyi, S., Yulong, Z., Yanzhen, J., Chen, C. & Liu, J. Micro interstitial fluid extraction and detection device integrated with the
optimal extraction conditions for noninvasive glucose monitoring. Biosens. Bioelectron. 237, 115515. https://​doi.​org/​10.​1016/j.​
bios.​2023.​115515 (2023).
22. Saraswathi, K., Sai Bhargava Reddy, M., Jayarambabu, N., Aich, S. & Venkatappa Rao, T. Highly sensitive non-enzymatic, non-
invasive disposable electrochemical polyaniline nanocaps based sweat sensor for glucose monitoring. Mater. Lett. 349, 134850.
https://​doi.​org/​10.​1016/j.​matlet.​2023.​134850 (2023).

Scientific Reports | (2024) 14:6151 | https://doi.org/10.1038/s41598-024-56677-z 10

Vol:.(1234567890)
www.nature.com/scientificreports/

23. Kiani, S., Rezaei, P. & Navaei, M. Dual-sensing and dual-frequency microwave SRR sensor for liquid samples permittivity detec-
tion. Measurement https://​doi.​org/​10.​1016/j.​measu​rement.​2020.​107805 (2020).
24. Feng, R. et al. Noninvasive blood glucose monitor via multi-sensor fusion and its clinical evaluation. Sens. Actuat. B Chem. 332,
129445. https://​doi.​org/​10.​1016/j.​snb.​2021.​129445 (2021).
25. Monte-Moreno, E. Non-invasive estimate of blood glucose and blood pressure from a photoplethysmograph by means of machine
learning techniques. Artif. Intell. Med. 53(2), 127–138. https://​doi.​org/​10.​1016/j.​artmed.​2011.​05.​001 (2011).
26. Gupta, H. S., Kwon, T.-H., Hossain, S. & Kim, K.-D. Towards non-invasive blood glucose measurement using machine learning:
An all-purpose PPG system design. Biomed. Signal Process. Control 68, 102706. https://d ​ oi.o
​ rg/1​ 0.1​ 016/j.b
​ spc.2​ 021.1​ 02706 (2021).
27. Han, G. et al. Noninvasive blood glucose sensing by near-infrared spectroscopy based on PLSR combines SAE deep neural network
approach. Infrared Phys. Technol. 113, 103620. https://​doi.​org/​10.​1016/j.​infra​red.​2020.​103620 (2021).
28. Yuchao, Fu., Huang, M. & Chen, X. Fingertip capillary dynamic near infrared spectrum (DNIRS) measurement combined with
multivariate linear modification algorithm for noninvasive blood glucose monitoring. Vib. Spectrosc. 113, 103223. https://​doi.​org/​
10.​1016/j.​vibsp​ec.​2021.​103223 (2021).
29. Guo, X. et al. Noninvasive in-vivo glucose detection in human finger interstitial fluid using wavelength-modulated differential
photothermal radiometry. J. Biophotonics https://​doi.​org/​10.​1002/​jbio.​20180​0441 (2019).
30. Rachim, V. P. & Chung, W.-Y. Wearable-band type visible-near infrared optical biosensor for non-invasive blood glucose monitor-
ing. Sens. Actuat. B Chem. 286, 173–180. https://​doi.​org/​10.​1016/j.​snb.​2019.​01.​121 (2019).
31. Diabetes | Healthcare Professionals | Abbott UK (n.d.). https://​www.​abbott.​co.​uk/​for-​profe​ssion​als/​diabe​tes-​care.​html (Accessed
1 May 2023).
32. Metasense (n.d.). https://​www.​oneca​re.​co.​in/​metas​ense (Accessed 1 May 2023).
33. How FreeStyle Works. The Reader (n.d.). https://​www.f​ reest​ yle.a​ bbott/i​ n-e​ n/h
​ ow-f​ reest​ yle-w​ orks-t​ herea​ der.h​ tml (Accessed 1 May
2023).

Author contributions
Abubeker K ­M1 *, Dr. R. R
­ amani2, and S B
­ askar5: Developed the concepts, design, experiments, validation and
manuscript preparation. Sreenivasulu ­Gogula3, Raja ­Krishnamoorthy4 and Kamalraj ­Subramaniam8: Concepts,
­ uthu6, Dr. Girinivasan ­Chellamuthu7:
Manuscript preparation, design validation, and proof reading. Dr. Sathish M
Concepts, device testing in real-time environment, and proof reading.

Competing interests
The authors declare no competing interests.

Additional information
Correspondence and requests for materials should be addressed to A.K.M.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/.

© The Author(s) 2024

Scientific Reports | (2024) 14:6151 | https://doi.org/10.1038/s41598-024-56677-z 11

Vol.:(0123456789)