The System of Automated Diabetes Control

Vitalii L. Levkivskyi

1 a

, Galyna V. Marchuk

1 b

, Oleksandr V. Kuzmenko

1 c

and Anton Yu. Levchenko

2 d

Zhytomyr Polytechnic State University, Department of Computer Science, 103 Chudnivska Str., Zhytomyr, 10003, Ukraine

Zhytomyr Polytechnic State University, Department of Software Engineering, 103 Chudnivska Str., Zhytomyr, 10003,

Keywords:

Diabetes, Disease, Glucose, Insulin, Glycemia, Data Analysis, Insulin/Glucose Balance.

Abstract:

Modern information technology creates new opportunities for quickly obtaining information about the status

and trends of diseases such as diabetes. This paper considers the task of automating the collection and analysis

of data on the condition of a patient. The object of the research is information technology for creating mobile

applications and mathematical models for calculating the optimal insulin therapy. The aim of the work is to

analyze the incidence of diabetes and the development of a software product for man-aging blood glucose

levels. We developed a self-control diary which is the main functional feature of the system. It allows users to

quickly obtain in-formation about the state and trends of the disease. Based on the accumulated daily data, it

is possible to take the necessary actions on time to improve the disease course. In the process of development,

the database was designed and mathematical methods were used to analyze the data. The module of the

analysis of the collected data for the speciﬁc period is developed. The results of the analysis can be obtained

by the following indicators: the average glucose blood level, the number of glucose measurements, glucose

deviations, the number of basal insulin injections, the number of hypo/hyperglycemia, the amount of bolus

insulin. In our work we used object-oriented design techniques, document-oriented databases, modern web

technology stacks for mobile application development and design of interfaces. The result of solving the

given task is a system of accumulation and systematization of statistical data on the course of the disease.

An automated diabetes control system has been designed and developed. Mathematical models were used to

calculate the glucose/insulin balance. The developed software can signiﬁcantly improve the living standards

of people with this disease. The system was tested by patients with diabetes.

1 INTRODUCTION

Diabetes mellitus, commonly known as diabetes, is

a chronic disease characterized by high blood glu-

cose level over a prolonged period of time. This

leads to serious problems in various systems of the

human body, especially nerve endings and blood ves-

sels. Diabetes is a dangerous complication that leads

to disability. In low- and middle-income countries,

the prevalence of diabetes is growing faster than in

high-income countries.

Diabetes is one of the leading causes of blindness,

kidney failure, heart attacks, strokes and lower ex-

tremity amputations. From 2000 to 2016, premature

https://orcid.org/0000-0002-1643-0895

https://orcid.org/0000-0003-2954-1057

https://orcid.org/0000-0002-4937-3284

https://orcid.org/0000-0002-4411-6465

mortality from diabetes increased by 5%. In 2019, di-

abetes be-came the ninth leading cause of death in the

world and is estimated to be the direct cause of 1.5

million deaths. According to the World Health Orga-

nization (WHO), the disease increases mortality by 2-

3 times and signiﬁcantly reduces life expectancy. At

the same time, the number of patients increases annu-

ally in all countries by 5-7%, and doubles every 12-15

years (World Health Organization, 2022).

Diabetes is treatable. A healthy diet, regular phys-

ical activity, maintaining a healthy weight and ab-

staining from tobacco use can prevent or delay the on-

set of diabetes. The implementation of the mobile ap-

plication ”Automated Diabetes Control System” will

allow for more effective treatment, thereby improving

the course of the disease. Users will be able to keep

a diary of self-control, particularly to enter and edit

data of physical activity, taking medication and food;

view disease-based analytics based on input and sync

Levkivskyi, V., Marchuk, G., Kuzmenko, O. and Levchenko, A.

The System of Automated Diabetes Control.

DOI: 10.5220/0012009500003561

In Proceedings of the 5th Workshop for Young Scientists in Computer Science and Software Engineering (CSSE@SW 2022), pages 41-49

ISBN: 978-989-758-653-8; ISSN: 2975-9471

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

data with an application server.

2 PROBLEM STATEMENT

Diabetes is a disease whose main symptom is a con-

stant rise of blood sugar level. In the human body, the

pancreas is responsible for stabilizing blood glucose

levels, producing the hormones insulin and glucagon,

which in-crease or decrease glucose levels. However,

patients with diabetes have pancreas malfunction in

combination with low insulin sensitivity. This leads

to ﬂuctuations in blood glucose levels, in particular to

the appearance of both hyperglycemia (high glucose

concentration) and hypoglycemia (low glucose con-

centration). Therefore, there is a need to ﬁnd treat-

ments that can improve the living standards of people

with this disease.

The main purpose of our research is to analyze

the problem of diabetes and to develop an automated

control system. The deﬁned purpose determines the

following tasks:

• to deﬁne the basic metrics and ways of their re-

ception for the development of the mathematical

module of the system;

• to design the structural components and algo-

rithms of the system;

• to develop an automated disease control applica-

tion.

3 REVIEW OF THE LITERATURE

The article “Global and regional diabetes prevalence

estimates for 2019 and projections for 2030 and 2045:

results from the International Diabetes Federation Di-

abetes Atlas, 9th edition” evaluates the prevalence of

diabetes in 2019 and forecasts for 2030 and 2045. The

study was conducted on 255 qualitative data sources

from 138 countries. Data was taken from adults aged

20-79 years for the period from 1990 to 2018, the

forecasts are not comforting (Saeedi et al., 2019). Ac-

cording to the IDF diabetes Atlas, the global preva-

lence of diabetes in 2021 is estimated at 10.5% (536.6

million people) and may increase to 12.2% (783.2

million people) in 2045 (Sun et al., 2022).

Saeedi et al. (Saeedi et al., 2020) estimated the

number of deaths related to diabetes among adults

aged 20–79. Diabetes is estimated to cause 11.3%

of deaths worldwide. Using a model that has only

one glucose compartment in its structure, the authors

conducted a number of simulations: taking into ac-

count the peculiarities of glucose absorption from the

intestine, they improved the procedure for detecting

latent forms of diabetes and analyzed the optimal in-

sulin therapy for an automated dispenser (Lapta et al.,

2014).

Levkivskyi et al. (Levkivskyi et al., 2020) are in-

vestigated the algorithms of data mining, which on

the basis of rules and calculations allow the creation

of a model that analyzes data by searching for cer-

tain patterns and trends. Through the study of data

mining algorithms, models and methods have been

developed to determine the impact of some chronic

diseases on others. The developed methods were im-

plemented in the system of intelligent data process-

ing. The conducted research testiﬁes to the prospects

of using methods of data mining to improve the qual-

ity of medical care for patients.

The research conducted by Bolodurina et al.

(Bolodurina et al., 2020) aims to develop and numeri-

cally solve the problem of optimal glycemic control

in patients with type 1 diabetes mellitus by insulin

therapy based on the conditions of optimality for non-

smooth systems with constant delay in the phase vari-

able.

In the article by Karpel’ev et al. (Karpel’ev et al.,

2015) the basic mathematical models of the biological

control system of plasma glucose concentration are

presented.

Palumbo et al. (Palumbo et al., 2013) offers a

method focused on the most important clinical / ex-

perimental tests conducted to understand the mecha-

nism of glucose homeostasis.

The dynamic behavior of a mathematical model,

conﬁrmed by experimental data, is studied by Trobia

et al. (Trobia et al., 2022) which takes into account

the relationship between glucose and insulin concen-

trations.

Shabestari et al. (Shabestari et al., 2018) presented

a new mathematical model to describe the interac-

tion between glucose, insulin and β-cells. The results

showed that the system shows different behaviors un-

der different conditions and is able to explain the in-

teraction between glucose, insulin and β-cells.

4 MATERIALS AND METHODS

4.1 Methods and Metrics for Modeling

an Automated Diabetes Control

System

To develop a system of automated control of diabetes,

it is necessary to deﬁne and implement methods for

collecting input data on the patient’s condition in real

CSSE@SW 2022 - 5th Workshop for Young Scientists in Computer Science Software Engineering

time. These data can be divided into two groups:

streaming and constant. Streaming data require reg-

ular input and include blood glucose, alcohol, carbo-

hydrates, weight, insulin, time and date, place of ad-

ministration, physical activity, stress and illness. The

constant data include such indicators as age, gender,

type of diabetes.

Some of these data can only be obtained by man-

ually entering it by a patient, others can be obtained

automatically. Information about blood glucose can

be obtained in two ways:

• after measuring blood glucose with a glucome-

ter. A patient can either enter data manually or

synchronize data with the system using Bluetooth

technology (if supported by the meter);

• with a continuous glucose monitoring system.

The monitor measures blood sugar every 10 sec-

onds and records the average value every 5 min-

utes.

After measuring the weight, the patient can either

enter the data or synchronize the data with the sys-

tem using Bluetooth technology (if supported by the

scale).

Information about physical activity has a differ-

ent nature of collection, one of which is the synchro-

nization of data from an external device such as a ﬁt-

ness tracker – a gadget that, in most cases, is worn on

the hand and has built-in sensors that monitor activity

during the day, including: number of steps, heart rate,

sleep, calories burned, etc. An analogue of a ﬁtness

tracker is data collection using an application installed

on the user’s smartphone, or the user can simply enter

data about daily activity manually.

If the data is entered manually, also the time and

date are required to be entered. If the data comes from

other devices or applications, it already contains time

and date information.

For developing the mathematical module of the

system metrics and methods for obtaining them were

determined.

Sokol et al. (Sokol et al., 2014) proposed the prin-

ciple of applying mathematical modeling to calculate

optimal insulin therapy. Bhonsle and Saxena (Bhon-

sle and Saxena, 2020) analyze various mathematical

models – despite the large number of mathematical

models, they are all based on one of two original

basic models: the model of the oral glucose toler-

ance test (OGTT) developed by Beaulieu in 1961 and

the model of the intravenous glucose tolerance test

(IGTT) by Bergman-Kobelli (Bergman et al., 1979).

The Beaulieu model is narrow in use, in particular,

it is generally unsuitable for describing the exponen-

tial decline of the glycemic curve of IGTT. The main

disadvantage of the IGTT model, in contrast to the

Beaulieu model, is that insulin is an input variable,

the value of which is determined clinically.

The mathematical model proposed by Shirokova

and Shirokov (Shirokova and Shirokov, 2006) is

based on the ratio of glucose balance and insulin con-

centration in human blood over a certain period of

time and improved by Bolodurina et al. (Bolodurina

et al., 2020).

Therefore, the experimental determination of

glycemic characteristics of insulin is as follows:

knowing the initial level of glucose in the blood, as

well as its integral characteristics, it is possible to

choose the right amount of insulin.

4.2 Design and Developing the

Algorithms of the System

For design and developing the automated diabetes

control system functional requirements were deﬁned:

• User registration and authentication must be pro-

vided in the system;

• Data storage: the system must store information

and allow the user to manage it;

• Keeping a diary of self-control: entering and edit-

ing data of physical activity, medication and food;

• Analytics review: the system should provide the

ability to review analytics for the selected period.

The algorithm of the automated diabetes control

system is presented in ﬁgure 1. First, the user gets

to the authorization screen, if he is registered, he can

immediately pass it and get to the main screen. Oth-

erwise, it is necessary to go through the registration

process by ﬁlling out a standard form, the entered data

are checked for validity and entered into the appropri-

ate collection in the database.

Once on the home screen, the user can immedi-

ately view the statistics. The user can also go to

the screen with analytics, where the information for

a speciﬁed period is displayed. To work with entries,

the diary screen with the functions of viewing, adding,

deleting and editing entries is available. It is possible

to set user settings, medication and glucose level. The

data entered by the user when making changes to the

settings or when working with diary entries are en-

tered into the database.

On the main screen it is possible to add a new di-

ary entry by clicking on the correspondent button. On

the opened modal window with a form current time

and date are passed. After the user enters the data, val-

idation and synchronization with the database are per-

formed and the user is redirected to the home screen

The System of Automated Diabetes Control

Figure 1: Activity diagram.

and methods are called to update the statistics based

on the entered data.

The analysis of functional requirements allowed

us to identify the following entities of the developing

process. Figure 2 shows a class diagram.

Some classes shown on the diagram are described

below.

• Home – a class of the main screen of the applica-

tion. It has the following methods: ionViewDid-

Load – this method starts after loading the screen

and initializes the methods for updating statis-

tics, updateStats – updating statistics for the week,

dailyStats – updating statistics for the day, pre-

pareStats – preparing statistics for the week, ﬁlter-

ByDate – ﬁltering diary entries by date, newNote

– create a new diary entry, getFormatedDate, get-

Time – get the date and time, showHelp – display

help information.

• Diary – a class for work with the screen of the

system. It has the following methods: ionView-

DidLoad – this method is started after loading the

screen and initializes the methods for updating the

list of entries, editNote – editing the entry.

• NewNotePage – a class for work with the modal

window of adding and editing entries. Its meth-

ods – showDatePicker, showTimePicker – display

forms for selecting the date and time, add, edit and

delete for making corresponding operations with

entries.

• Analytics – a class for work with the analyt-

ics screen. It has the following methods: ion-

ViewDidLoad – it is called after loading the

screen and initializes methods for analytics up-

date, updateAnalytics, prepareReport (analytics

data preparation), ﬁlterByDate (ﬁltering entries

by date), getFormatedDate, setFormatedDate,

showDatePicker (getting and setting dates, date

selection form correspondingly).

• Settings – a class for work with settings.

• Insulin-settings and Glucose-settings are classes

for work with medicine and glucose settings.

They have the following methods: ionViewDid-

Load – the method is called after loading the

screen and loads the settings, ionViewWillLeave –

the method is called before closing the screen and

saves the changes in the settings.

• Personal-settings – a class for work with per-

sonal set-tings. The class has the following meth-

ods: ionViewDidLoad, ionViewWillLeave, show-

DatePicker.

• Login is a class for the user authentication screen.

It has a method login for user authentication.

• Register is a class for the user registration screen.

It has a method register for new user registration.

• SettingsProvider – a class for working with user

set-tings in the database. Its methods: getSet-

tings – to get set-tings from the database, update-

Settings – to update the settings in the database,

addNewUserSettings – to create a document with

the settings of a new user.

• NotesProvider – a class for working with user di-

ary entries in the database. Its methods are get-

Notes – to load all user entries from the database,

addNewNote – to add a new entry to the database,

editNote – to update the entry in the database,

deleteNote – to delete the entry from the database.

CSSE@SW 2022 - 5th Workshop for Young Scientists in Computer Science Software Engineering

Figure 2: Class diagram.

The System of Automated Diabetes Control

Thus, the developed application has the function-

ality of making and editing entries in the system,

medicine selection, obtaining analytics, data synchro-

nization between devices, provided by the methods

of the components Home, Diary, NewNotePage, An-

alytics, Settings, Login, Register, Personal-settings,

Insulin-settings, Glucose-settings, SettingsProvider,

NotesProvider.

Functions in the class Analytics are used to per-

form analytics. The code snippet is below:

updateAnalytics() {

this.notesForAnalytics =

this.notesSortedByDate.filter(note =>

this.filterByDate(note,this.dateFrom,

this.dateTo));

let report;

let collectedData = {

noteCounter: 0,

glucoseCounter: 0,

glucoseSum: 0,

hiLowCounter: 0,

bolusInjectionCounter: 0,

basalUCounter: 0,

bolusUCounter: 0

}

this.notesForAnalytics.forEach(function

(noteScope, i, notes) {

noteScope.stats.forEach(function(note, i,

notesFromScope){

collectedData.noteCounter++;

if(note.stats.glucose){

collectedData.glucoseCounter++;

collectedData.glucoseSum +=

parseFloat(note.stats.glucose);

if(note.stats.glucose <

this.glucoseSettings.lowLevel

|| note.stats.glucose >

this.glucoseSettings.hiLevel){

collectedData.hiLowCounter++;

}

if(note.stats.bolus){

collectedData.bolusInjectionCounter++;

collectedData.bolusUCounter +=

parseFloat(note.stats.bolus);

}

if(note.stats.basal){

collectedData.basalUCounter +=

parseFloat(note.stats.basal);

}

}.bind(this));

report = this.prepareReport(collectedData);

if (i === 0){

if (noteScope.day === this.day) {

this.dayReport = report;

}

this.weekReport = report;

this.monthReport = report;

} else if ( i <= 6){

this.weekReport = report;

this.monthReport = report;

} else {

this.monthReport = report;

}

}.bind(this)); }

To store and access system information a database

was designed. It consists of three main collections

(User, UserSettings, UserNotes) (ﬁgure 3).

5 RESULTS

After starting the automated diabetes control system,

the login screen is shown, where the user must lo-

gin or register by clicking on the registration button.

After successful authentication the main screen is dis-

played (ﬁgure 4) with statistics of the main indicators

for the week and for the current day, namely: bolus

insulin per day, the amount of active bolus insulin at

the moment, the number of hypo/hyperglycemia for

the current day, average sugar level in seven days,

mean deviation of sugar in seven days and the amount

of hypo/hyperglycemia in seven days in percent. By

clicking on these indicators, the user can get reference

information (ﬁgure 4).

On the main screen there is a button to create

a new diary entry. By clicking on it a modal win-

dow with a form is opened. Navigation is carried

out through four tabs: Home screen, Diary, Analyt-

ics, Settings. On the screen of a diary (ﬁgure 5), all

entries sorted and grouped by date and time are dis-

played as rows of cards with entered metrics.

By clicking on any entry, a modal window is

opened to edit or delete the entry (ﬁgure 6). This

window has a form with all the necessary ﬁelds for

ﬁlling: medicine, food, activity, sugar level. The user

can also choose the time and date of the entry. In

editing mode of an entry, a button for deleting an en-

try appears in the navigation bar with a dialog box to

conﬁrm the deletion of the record.

On the screen of analytics the user can choose the

desired period of a report, namely: one day, seven

days and thirty days. Also, the date can be speciﬁed

from which the data will be calculated (ﬁgure 7).

The screen with settings displays the user’s email

and a list of settings groups: personal settings,

glucose settings for adjusting glucose levels for

hypo/hyperglycemia, medication settings for choos-

ing medications. There is also a login button in the

navigation bar.

CSSE@SW 2022 - 5th Workshop for Young Scientists in Computer Science Software Engineering

Figure 3: Database schema.

Figure 4: Home screen. Reference information.

Figure 5: The screen of a diary.

The System of Automated Diabetes Control

Figure 6: The forms for creating and editing entries.

Figure 7: The screen of analytics.

6 CONCLUSIONS

It is estimated that diabetes is the cause of one in nine

deaths among adults aged 20-79. Prevention of dia-

betes and its complications is important, especially in

middle-income countries, where the current impact is

estimated to be greatest (Saeedi et al., 2020).

As a result of the research, an analysis of the prob-

lem of diabetes was conducted. The basic metrics

and methods of production of these metrics for the

mathematical module of the system are determined.

The functional requirements for the automated dis-

ease control system were analyzed. Algorithms of the

system functions were deﬁned and described, the or-

der of interaction of classes during the execution of

the programing code was determined and the system

of automated control of diabetes mellitus was devel-

oped.

In the process of development and testing, we con-

sulted with doctors and took into account their recom-

mendations. The system has received positive feed-

back from diabetes patients who continue to use it.

The developed software product is ready for use.

REFERENCES

Bergman, R. N., Ider, Y. Z., Bowden, C. R., and Cobelli, C.

(1979). Quantitative estimation of insulin sensitivity.

American Journal of Physiology-Endocrinology and

Metabolism, 236(6):E667. https://doi.org/10.1152/

ajpendo.1979.236.6.E667.

Bhonsle, S. and Saxena, S. (2020). A review on control-

relevant glucose–insulin dynamics models and regu-

lation strategies. Proceedings of the Institution of Me-

chanical Engineers, Part I: Journal of Systems and

Control Engineering, 234(5):596–608. https://doi.org/

10.1177/0959651819870328.

Bolodurina, I. P., Ivanova (Lugovskova), Y. P., and Antsif-

erova, L. M. (2020). Optimal Control of Glycemia

Regulation Dynamics in Patients with Type I Diabetes

Mellitus. Bulletin of the South Ural State Univer-

sity. Ser. Computer Technologies, Automatic Control,

Radio Electronics, 20(4):144–154. https://doi.org/10.

14529/ctcr200415.

Karpel’ev, V. A., Filippov, Y. I., Tarasov, Y. V., Boyarsky,

M. D., Mayorov, A. Y., Shestakova, M. V., and Dedov,

I. I. (2015). Mathematical Modeling of the Blood Glu-

cose Regulation System in Diabetes Mellitus Patients.

Annals of the Russian academy of medical sciences,

70(5):549–560. https://doi.org/10.15690/vramn.v70.

i5.1441.

Lapta, S. S., Pospelov, L. A., and Solovieva, O. (2014).

Computerized early diagnosis of diabetes mellitus by

methods of mathematical modeling. Vestnik NTU

KHPI, 36(1079):55–61. http://repository.kpi.kharkov.

ua/handle/KhPI-Press/9388.

Levkivskyi, V., Lobanchykova, N., and Marchuk, D.

(2020). Research of algorithms of Data Mining. E3S

Web of Conferences, 166:05007. https://doi.org/10.

1051/e3sconf/202016605007.

Palumbo, P., Ditlevsen, S., Bertuzzi, A., and De Gae-

tano, A. (2013). Mathematical modeling of the

glucose–insulin system: A review. Mathematical

Biosciences, 244(2):69–81. https://doi.org/10.1016/j.

mbs.2013.05.006.

Saeedi, P., Petersohn, I., Salpea, P., Malanda, B., Karu-

ranga, S., Unwin, N., Colagiuri, S., Guariguata, L.,

Motala, A. A., Ogurtsova, K., Shaw, J. E., Bright, D.,

and Williams, R. (2019). Global and regional dia-

betes prevalence estimates for 2019 and projections

for 2030 and 2045: Results from the International

Diabetes Federation Diabetes Atlas, 9

edition. Di-

abetes Research and Clinical Practice, 157:107843.

https://doi.org/10.1016/j.diabres.2019.107843.

CSSE@SW 2022 - 5th Workshop for Young Scientists in Computer Science Software Engineering

Saeedi, P., Salpea, P., Karuranga, S., Petersohn, I., Ma-

landa, B., Gregg, E. W., Unwin, N., Wild, S. H., and

Williams, R. (2020). Mortality attributable to dia-

betes in 20–79 years old adults, 2019 estimates: Re-

sults from the International Diabetes Federation Dia-

betes Atlas, 9

edition. Diabetes Research and Clin-

ical Practice, 162:108086. https://doi.org/10.1016/j.

diabres.2020.108086.

Shabestari, P. S., Panahi, S., Hatef, B., Jafari, S., and

Sprott, J. C. (2018). A new chaotic model for glucose-

insulin regulatory system. Chaos, Solitons & Frac-

tals, 112:44–51. https://doi.org/10.1016/j.chaos.2021.

111753.

Shirokova, N. A. and Shirokov, I. V. (2006). Mathematical

model of the balance “glucose – insulin – glucagon” in

human blood. Bulletin of Omsk University, (3):51–53.

https://tinyurl.com/5dsampk5.

Sokol, E. I., Lapta, S. S., Pospelov, L. A., and Solovieva,

O. I. (2014). Raschet rezhimov insulinoterapii

na osnove ma-tematicheskogo komp’juternogo mod-

elirovanija. Visnik NTU ”HPI”, 36(1079):61–66. http:

//repository.kpi.kharkov.ua/handle/KhPI-Press/9389.

Sun, H., Saeedi, P., Karuranga, S., Pinkepank, M.,

Ogurtsova, K., Duncan, B. B., Stein, C., Basit, A.,

Chan, J. C., Mbanya, J. C., Pavkov, M. E., Ramachan-

daran, A., Wild, S. H., James, S., Herman, W. H.,

Zhang, P., Bommer, C., Kuo, S., Boyko, E. J., and

Magliano, D. J. (2022). IDF Diabetes Atlas: Global,

regional and country-level diabetes prevalence esti-

mates for 2021 and projections for 2045. Diabetes

Research and Clinical Practice, 183:109119. https:

//doi.org/10.1016/j.diabres.2021.109119.

Trobia, J., de Souza, S. L. T., dos Santos, M. A., Szezech,

J. D., Batista, A. M., Borges, R. R., da S. Pereira,

L., Protachevicz, P. R., Caldas, I. L., and Iarosz,

K. C. (2022). On the dynamical behaviour of a

glucose-insulin model. Chaos, Solitons & Frac-

tals, 155:111753. https://doi.org/10.1016/j.chaos.

2021.111753.

World Health Organization (2022). Diabetes. https://www.

who.int/news-room/fact-sheets/detail/diabetes.

The System of Automated Diabetes Control