698
Resea chBib IF - 11.01, ISSN: 3030-3753, Volume 2 Issue 11
УДК 618.3-07:004.8
PREDICTION OF PREECLAMPSIA DEVELOPMENT USING ARTIFICIAL
INTELLIGENCE
Nema o a Ma jona Zik illae na
[email p o ec ed]
h ps://o cid.o g/0009-0000-4105-1064
Bukha a S a e Medical Ins i u e named a e Abu Ali Ibn Sina, Bukha a, Uzbekis an.
h ps://doi.o g/10.5281/zenodo.17688397
Abs ac . P eeclampsia emains one o he leading causes o ma e nal and pe ina al
mo ali y wo ldwide. Timely p edic ion o his p egnancy complica ion signi ican ly educes he
isk o se e e ou comes o bo h he mo he and he e us. T adi ional diagnos ic me hods, based
on clinical and labo a o y indica o s, o en de ec he pa hology a la e s ages. The e o e, he
use o a i icial in elligence (AI) echnologies o ea ly p edic ion o p eeclampsia de elopmen ,
based on big da a and mul i ac o ial analysis, has gained pa icula impo ance.
Keywo ds: p eeclampsia, p egnancy, p edic ion, a i icial in elligence, machine lea ning,
XGBoos , complica ion p e en ion.
ПРОГНОЗИРОВАНИЕ РАЗВИТИЯ ПРЕЭКЛАМПСИИ С ИСПОЛЬЗОВАНИЕМ
ИСКУССТВЕННОГО ИНТЕЛЛЕКТА
Аннотация. Преэклампсия остаётся одной из ведущих причин материнской и
перинатальной смертности во всём мире. Своевременное прогнозирование данного
осложнения беременности позволяет значительно снизить риск тяжёлых исходов для
матери и плода. Традиционные методы диагностики, основанные на клинико-
лабораторных показателях, нередко выявляют патологию уже на поздних стадиях. В
связи с этим особое значение приобретает использование технологий искусственного
интеллекта (ИИ) для раннего прогнозирования развития преэклампсии на основе больших
данных и мультифакторного анализа.
Ключевые слова: преэклампсия, беременность, прогнозирование, искусственный
интеллект, машинное обучение, XGBoos , профилактика осложнений.
SUN’IY INTELLEKT YORDAMIDA PREEKLAMPSIYA RIVOJLANISHINI
BASHORATLASH
Anno a siya. P eeklampsiya dunyo bo‘ylab ona a pe ina al o‘limning ye akchi
sababla dan bi i bo‘lib qolmoqda. Ushbu homilado lik aso a ini o‘z aq ida oldindan aniqlash
ona a homila uchun og‘i oqiba la xa ini sezila li da ajada kamay i adi. An’ana iy
diagnos ika usulla i, ya’ni klinik a labo a o iya ko‘ sa kichla iga asoslangan usulla ,
ko‘pincha pa ologiyani aqa kech bosqichla da aniqlaydi. Shu sababli, ka a ma’lumo la a
ko‘p omilli ahlil asosida p eeklampsiya i ojlanishini e a p ognoz qilishda sun’iy in ellek (SI)
exnologiyala idan oydalanish muhim ahamiya kasb e moqda.
Kali so‘zla : p eeklampsiya, homilado lik, p ognozlash, sun’iy in ellek , mashina
o‘ ganish, XGBoos , aso a la ni oldini olish.
In oduc ion
P eeclampsia is a mul i ac o ial p egnancy-speci ic diso de cha ac e ized by
hype ension and p o einu ia ha ypically de elops a e 20 weeks o ges a ion [1]. I emains
one o he leading causes o ma e nal and pe ina al mo bidi y and mo ali y wo ldwide.
699
Resea chBib IF - 11.01, ISSN: 3030-3753, Volume 2 Issue 11
Acco ding o he Wo ld Heal h O ganiza ion (WHO), p eeclampsia complica es
app oxima ely 5–8% o all p egnancies globally, con ibu ing o 10–15% o ma e nal dea hs and
20–25% o pe ina al dea hs each yea [2,3]. The condi ion a ec s a ound 8.5 million women
annually, wi h he highes bu den obse ed in low- and middle-income coun ies, whe e access
o imely diagnos ic and p e en i e ca e emains limi ed [4].
The pa hophysiology o p eeclampsia is complex and no ye ully unde s ood. I is
hough o esul om abno mal placen al de elopmen , impai ed ophoblas ic in asion,
endo helial dys unc ion, and sys emic in lamma o y esponses ha lead o mul isys em ma e nal
o gan in ol emen [5]. The unp edic able and apid p og ession o p eeclampsia makes ea ly
diagnosis and isk p edic ion c ucial o p e en ing se e e ma e nal and e al ou comes, such as
eclampsia, placen al ab up ion, p e e m bi h, and in au e ine g ow h es ic ion [6,7].
T adi ional sc eening app oaches, which ely mainly on ma e nal his o y, clinical
e alua ion, and isola ed biochemical ma ke s (such as se um placen al g ow h ac o o mean
a e ial p essu e), o en show limi ed sensi i i y and speci ici y [8]. These con en ional me hods
a e unable o cap u e he complex in e play o biological, en i onmen al, and gene ic ac o s ha
con ibu e o disease de elopmen [9].
In ecen yea s, he g owing a ailabili y o big da a in obs e ics and ad ances in a i icial
in elligence (AI) and machine lea ning (ML) echnologies ha e opened new oppo uni ies o
p edic i e modeling. AI-based algo i hms can in eg a e la ge, mul idimensional da ase s
including demog aphic, clinical, labo a o y, and imaging pa ame e s and au oma ically de ec
sub le, nonlinea associa ions ha may no be e iden o human analysis. S udies conduc ed in
Eu ope, he Uni ed S a es, and Asia ha e demons a ed ha machine lea ning models can
achie e up o 90–95% accu acy in p edic ing p eeclampsia isk a ea ly ges a ional s ages [10].
The in eg a ion o a i icial in elligence in o obs e ic p ac ice hus ep esen s a
ans o ma i e s ep owa d pe sonalized, da a-d i en ma e nal heal hca e. Ea ly iden i ica ion o
high- isk pa ien s using AI-powe ed p edic ion ools may allow imely in e en ion, close
moni o ing, and he implemen a ion o p e en i e s a egies such as low-dose aspi in he apy.
The e o e, his s udy aims o de elop and e alua e an AI-based p edic i e model o he
ea ly iden i ica ion o women a isk o p eeclampsia, combining clinical, biochemical, and
demog aphic da a o imp o e he accu acy o p edic ion and con ibu e o educing he global
bu den o ma e nal and neona al mo bidi y and mo ali y.
Aim o he S udy. To de elop and e alua e he e ec i eness o a p edic i e model o
p eeclampsia using machine lea ning algo i hms.
Ma e ials and Me hods
This e ospec i e analy ical s udy was conduc ed on a coho o 1,200 p egnan women
who ecei ed an ena al ca e a h ee egional obs e ic hospi als be ween 2020 and 2024. The
selec ion c i e ia included women wi h single on p egnancies be ween 10 and 20 weeks o
ges a ion and comple e clinical and labo a o y da a. Pa ien s wi h p e-exis ing enal disease,
au oimmune diso de s, o mul iple p egnancies we e excluded om he analysis. The diagnosis
o p eeclampsia was made acco ding o he c i e ia o he Ame ican College o Obs e icians and
Gynecologis s (ACOG), which de ine he condi ion as ele a ed blood p essu e (≥140/90 mmHg)
on wo o mo e occasions a e 20 weeks o ges a ion, accompanied by p o einu ia o o he signs
o ma e nal o gan dys unc ion.
Da a we e ob ained om medical eco ds and included demog aphic, clinical,
biochemical, and Dopple ul asound indica o s.
700
Resea chBib IF - 11.01, ISSN: 3030-3753, Volume 2 Issue 11
The analyzed a iables comp ised ma e nal age, body mass index (BMI), pa i y, amily
his o y o hype ension o p eeclampsia, mean a e ial p essu e, ch onic hype ension, diabe es
melli us, and biochemical pa ame e s such as se um placen al g ow h ac o (PlGF), soluble ms-
like y osine kinase-1 (sFl -1), u ic acid, and C- eac i e p o ein (CRP).
In addi ion, u e ine a e y pulsa ili y and esis ance indices we e e alua ed by Dopple
ul asonog aphy. In o al, 35 pa ame e s we e included o model aining and alida ion.
Resul s and Discussion
The de eloped a i icial in elligence–based models demons a ed a high le el o accu acy
in p edic ing he isk o p eeclampsia. Among he es ed algo i hms, he Ex eme G adien
Boos ing (XGBoos ) model achie ed he bes o e all pe o mance, wi h an a ea unde he ROC
cu e (AUC) o 0.94, sensi i i y o 91%, speci ici y o 89%, and an o e all p edic ion accu acy
o 90%.
The andom o es and suppo ec o machine models showed sligh ly lowe
pe o mance, wi h AUC alues o 0.89 and 0.87, espec i ely, while logis ic eg ession eached
0.82. These indings con i m he supe io i y o ensemble and g adien boos ing me hods in
handling complex, nonlinea ela ionships be ween mul iple clinical and biochemical p edic o s.
Fea u e impo ance analysis iden i ied se e al key a iables con ibu ing mos
signi ican ly o he p edic ion o p eeclampsia. The leading p edic o s included mean a e ial
p essu e du ing he second imes e , ma e nal age o e 35 yea s, he u e ine a e y pulsa ili y
index, and biochemical ma ke s such as placen al g ow h ac o (PlGF) and soluble ms-like
y osine kinase-1 (sFl -1).
These indings align wi h p e ious s udies indica ing ha endo helial dys unc ion and
impai ed placen al pe usion a e he cen al mechanisms unde lying he de elopmen o
p eeclampsia. Ele a ed sFl -1 and educed PlGF concen a ions ha e been ecognized as ea ly
bioma ke s e lec ing placen al ischemia and endo helial ac i a ion.
The applica ion o AI algo i hms allowed he in eg a ion o 35 di e se clinical,
demog aphic, and biochemical ea u es, p oducing a comp ehensi e isk e alua ion o each
pa icipan . The model’s high p edic i e pe o mance demons a es he easibili y o using AI-
based ools in clinical obs e ics o iden i y women a high isk be o e he onse o clinical
symp oms.
This could enable a ge ed p e en i e measu es such as low-dose aspi in he apy, mo e
equen blood p essu e moni o ing, and close e al su eillance, ul ima ely imp o ing ma e nal
and neona al ou comes. Ou esul s a e consis en wi h indings om in e na ional s udies, which
epo ed simila p edic i e accu acy le els (AUC 0.90–0.95) when applying AI and machine
lea ning echniques o p eeclampsia isk assessmen .
The abili y o AI sys ems o p ocess la ge olumes o he e ogeneous da a, including
nonlinea ela ionships ha a e di icul o cap u e h ough adi ional s a is ical models,
unde sco es hei impo ance in mode n pe ina al medicine.
701
Resea chBib IF - 11.01, ISSN: 3030-3753, Volume 2 Issue 11
Figu e 1. Compa a i e pe o mance o machine lea ning models o p edic ing
p eeclampsia based on ROC-AUC sco es.
O e all, he indings con i m ha a i icial in elligence can se e as a powe ul
ins umen o ea ly p edic ion o p eeclampsia, allowing clinicians o adop a pe sonalized and
p e en i e app oach o ma e nal ca e. The use o AI-d i en p edic i e models ep esen s a
signi ican ad ancemen owa d p ecision obs e ics and may con ibu e o educing he global
bu den o hype ensi e diso de s o p egnancy.
Conclusion
The p esen s udy demons a ed he easibili y and e ec i eness o using a i icial
in elligence–based app oaches o p edic ing he de elopmen o p eeclampsia in p egnan
women. Among he es ed models, he XGBoos algo i hm achie ed he highes p edic i e
accu acy, sensi i i y, and speci ici y, con i ming i s sui abili y o clinical implemen a ion. The
in eg a ion o clinical, biochemical, and Dopple pa ame e s in o a single p edic i e model
allowed o ea ly iden i ica ion o women a high isk o p eeclampsia, e en be o e he onse o
clinical symp oms.
The indings indica e ha AI-d i en p edic ion sys ems can signi ican ly enhance he
p ecision o obs e ic isk assessmen compa ed o adi ional s a is ical me hods. By de ec ing
complex nonlinea ela ionships among mul iple isk ac o s, machine lea ning models p o ide a
powe ul ool o pe sonalized and p e en i e obs e ic ca e.
Implemen ing such models in clinical p ac ice could help obs e icians o s a i y
p egnan women acco ding o indi idual isk p o iles, ini ia e p e en i e in e en ions—such as
low-dose aspi in he apy o in ensi ied moni o ing and ul ima ely educe ma e nal and pe ina al
mo bidi y and mo ali y associa ed wi h p eeclampsia.
Fu u e esea ch should ocus on expanding he da ase o include mul i-cen e ,
p ospec i e da a and in eg a ing gene ic, me abolic, and en i onmen al ac o s o u he imp o e
p edic i e accu acy and gene alizabili y. O e all, he use o a i icial in elligence ep esen s a
majo ad ancemen owa d p ecision obs e ics, o e ing new possibili ies o ea ly de ec ion,
imely in e en ion, and imp o ed ou comes in ma e nal heal h ca e.
702
Resea chBib IF - 11.01, ISSN: 3030-3753, Volume 2 Issue 11
Li e a u e
1. Аюшева С.Э., Жданова М.С., Пономарева Е.А. Современные подходы к
прогнозированию и профилактике преэклампсии // Акушерство и гинекология. –
2022. – №5. – С. 45–52.
2. B own M.A., Magee L.A., Kenny L.C. e al. The hype ensi e diso de s o p egnancy:
ISSHP classi ica ion, diagnosis & managemen ecommenda ions o in e na ional
p ac ice // P egnancy Hype ension. – 2018. – Vol. 13. – P. 291–310.
3. Савельева Г.М., Курцер М.А., Шалина Р.И. Преэклампсия: современные аспекты
патогенеза, диагностики и терапии. – М.: ГЭОТАР-Медиа, 2021. – 328 с.
4. Rana S., Lemoine E., G ange J.P., Ka umanchi S.A. P eeclampsia: pa hophysiology,
challenges, and pe spec i es // Ci cula ion Resea ch. – 2019. – Vol. 124(7). – P. 1094–
1112.
5. Касымова Н.А., Халилова Г.Р., Ибрагимова Д.Ш. Роль факторов риска плода в
развитии осложнений беременности//Вестник репродуктивного здоровья. – 2023. –
№2. – С. 25–30.
6. Liu X., Chen M., Zhao J. e al. Machine lea ning-based p edic ion o p eeclampsia using
ma e nal and e al pa ame e s//F on ie s in Medicine. – 2021. – Vol. 8. – A icle 625.
7. Мухамедова З.Ш., Турсунова Г.Б., Хамраева Н.М. Использование технологий
искусственного интеллекта в перинатальной диагностике//Журнал клинической
медицины Узбекистана. – 2023. – №4. – С. 57–63.
8. Chappell L.C., Clu e C.A., Kingdom J., Tong S. P e-eclampsia // Lance . – 2021. – Vol.
398(10297). – P. 341–354.
9. Абдуллаева М.Р., Юсупова Д.А. Перспективы внедрения интеллектуальных систем
в прогнозировании акушерских осложнений // Медицинский вестник Бухары. –
2024. – №1. – С. 18–24.
10. Zhang Y., Wang H., Li Q. e al. A i icial in elligence-assis ed p edic ion o p eeclampsia
based on e al ul asound and ma e nal bioma ke s // BMC P egnancy and Childbi h. –
2022. – Vol. 22. – A icle 154.
