=Paper=
{{Paper
|id=Vol-3309/paper3
|storemode=property
|title=Predicting the risk of surgical complications based on multiparametric neural network clustering at revascularization of main arteries of the lower limbs
|pdfUrl=https://ceur-ws.org/Vol-3309/paper3.pdf
|volume=Vol-3309
|authors=Boryslav Selskyi,Sviatoslav Kostiv,Ihor Venher,Petro Selskyi
|dblpUrl=https://dblp.org/rec/conf/ittap/SelskyiKVS22
}}
==Predicting the risk of surgical complications based on multiparametric neural network clustering at revascularization of main arteries of the lower limbs==
https://ceur-ws.org/Vol-3309/paper3.pdf
Predicting the risk of surgical complications based on
multiparametric neural network clustering at revascularization
of main arteries of the lower limbs
Selskyi Boryslava, Sviatoslav Ya Kostiva, Ihor K.Venhera and Petro R Selskyia
a
I. Horbachevsky Ternopil National Medical University, Maidan Voli, 1, Ternopil, 46002, Ukraine
Abstract
To propose a method for predicting the complications of surgical interventions through the use
of multiparameter neural network clustering, followed by the development of a scale for
stratification of the surgical risk complications. Analysis of examination results of 411 patients
with obliterating atherosclerosis of the main arteries of the lower limbs was performed. For a
more in-depth analysis of the combined changes in the performance of the studied groups in
order to optimize the prediction of the risk of complications in the postoperative period, neural
network clustering was performed by using the software NeuroXL Classifier. The proposed
scale of the surgical risk stratification of treatment the main arteries of the lower extremities in
the postoperative period takes into account the multifactorial clinical-anamnestic and
laboratory-instrumental studies. Considering the combination of factors that characterize the
state of organs and systems, influence the choice and method of reconstructive surgery. Based
on the data of neural network clustering, the level of possible postoperative complications of
surgery on the main arteries of the lower extremities was determined, followed by four levels
of risk factors development: 31-40 very high risk, high risk 21-30, moderate 11-20, and low -
1–10.
Keywords 1
predicting the risk factors, neural network clustering, risk scale, obliterating atherosclerosis,
revascularization, vascular reconstruction.
1. Introduction
Application of information methods and computer modeling in modern conditions makes it possible
to significantly improve the quality and provide a comprehensive approach to the choice of surgical
intervention, especially in the field of vascular surgery, which requires a large amount of information
and parameters analysis. [1,5] A large number of studies have been aimed at solving prevention of
complications, taking into account patient’s parameters and developing relevant prevention risk scales
[7]. However, the problem of their usage in practical medicine and comprehensive consideration of
numerous risk factors is still being solved [8, 9]. At the same time, the development of a unified scale
of possible risks of surgical intervention has become extremely important. The prediction of combined
complications in patients with vascular pathology and the use of neural network technology for their
detection have remained especially applicable [10, 11].
ITTAP’2022: 2nd International Workshop on Information Technologies: Theoretical and Applied Problems, November 22–24, 2022, Ternopil,
Ukraine
EMAIL: selskyi_bp@tdmu.edu.ua l (A.1); kostivsj@tdmu.edu.ua (A.2); vengerik@tdmu.edu.ua (A.3); selskyy@tdmu.edu.ua (A.4)
ORCID: 0000-0001-6787-4843 (A.1); 0000-0002-7963-5425 (A. 2); 0000-0003-0170-1995 (A. 3); 0000-0001-9778-2499 (A.4)
©️ 2022 Copyright for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR Workshop Proceedings (CEUR-WS.org)
2. Main Part
To form a multi-parameter neural network clustering followed by the development of a risk
stratification scale for surgical complications, we have performed analysis of the examination indicators
of 411 patients with obliterating atherosclerosis of the main arteries of lower limbs.
To determine the nature and prevalence of atherosclerotic lesions of the arterial bed of major arteries
of lower limbs and to examine patiens, we used ultrasound SonoScape S8 Exp (Italy) and tomographic
computer study Siemens Brilliance CT64 (Germany) with contrasting of vascular bed. For angiography
of the vessels of the lower limbs in the conditions of the endovascular operating room with X-ray we
used angiograph Siemens Axiom Artis (Germany).
The obtained results of clinical observations, laboratory and instrumental studies have been
processed by variational mathematical statistics method. To process of statistical data, we used
Microsoft Excel (2013) package. In cases of normal distribution, statistical significance of the difference
between the arithmetic avarage was defined with the help of Student's test (T-Test), and in the case of
non-normal distribution – with the help of Mann-Whitney non-parametric test (U- Test) at p<0.05.
For a more in-depth analysis of combined changes in the indicators of groups in observational studies
and in order to optimize the prediction of the risk of developing complications in the postoperative
period, we have performed a neural network clustering using NeuroXL Classifier add-in for the
Microsoft Excel program. The NeuroXL Classifier program (developed by the AnalyzerXL company)
implements self-organizing neural networks that process categorization by studying trends and
interconnections within groups. The key advantages of NeuroXL Classifier are simplicity of mastering
and usage; in-depth knowledge in the field of neural networks is optional; integration with Microsoft
Excel; provision of justified neural network technology for high-accuracy classification; determination
of interconnections and trends that cannot be defined by traditional statistical methods [1, 2, 3, 4].
3. Results and Discussions
We have analyzed clinical observations results, laboratory and instrumental studies, which we
entered in the neural network clustering system for processing.
3.1. Average values of indicators
Besides, we have analyzed anamnestic, and clinical and laboratory indicators of 72 patients (group
1). Open surgical interventions (subgroup 1a) have been used to treat 44 patients (61.1%), and
endovascular interventions (subgroup 1b) – to treat 28 patients (38.9%). The average age of patients in
the first group was (67.06±1.14) years old. The body mass index (BMI) in patients of this examination
group was 23.39±0.39. (50.0±5.89)% of patients led unhealthy lifestyle. Among other indicators, the
following were taken into account: lesions of extracranial arteries ((54.14±5.87) %), diabetes
((30.56±5.43) %), stroke in history ((5.56±2.70 ) %), myocardial infarction in history ((23.61±5.01) %),
pathology of the gastrointestinal tract ((15.28±4.24) %), respiratory failure ((12.5±3.90 ) %), diseases
of the cardiovascular system ((95.83±2.35) %), malignant process in history ((1.39±1.38) %),
conduction anesthesia ((13.89±4.08 ) %), epidural anesthesia ((55.56±5.86) %), mechanical ventilation
((1.39±1.38) %), presence of pulmonary hypertension ((9.72±3.49) %) and level revascularization
((5.56±2.70) %).
(25±5.10)% of patients (subgroup 1c) suffered from side-effects, such as thrombosis of the
reconstruction segment ((19.44±4.66) %), myocardial infarction ((1.39±1.38) %), pseudoaneurysm
((2.78±1.94) %) and suppuration conduit ((4.17±2.35) %). It should be noted that the average age
((67.5±1.74) years old and BMI (22.64±0.89) of the patients in this subgroup were not significantly
different from the similar indicators of the patients of the first group of our study (р>0.05 ).
We have studied general blood analysis and biochemical indicators and coagulogram indicators of
all subgroups of patients. The results of the study are shown in Table 1.
Table 1
Indicators of general blood analysis, biochemical indicators and coagulogram indicators of patients with
open and endovascular surgical interventions (М m).
1 group 1а subgroup 1b subgroup 1c subgroup
Indicators
(n – 72) (n – 44) (n – 28) (n – 18)
Erythrocytes,
* 4,38±0,06 4,32±0,05 4,48±0,08 4,50±0,28
1012/l
Hemoglobin,
128,80±2,41 127,30±2,92 131,18±4,18 125,72±4,97
g/dl
Color index 0,90±0,01 0,91±0,01 0,89±0,01 0,84±0,05
Leukocytes, *109 /l 7,85±0,39 8,16±0,54 7,35±0,54 7,23±0,64
Eosinophils, % 3,26±0,76 2,73±0,36 4,09±1,87 2,72±0,48
Rod-shaped
6,58±0,50 6,75±0,73 6,32±0,60 6,22±0,77
neutrophils, %
Segmented
64,58±1,11 65,66±1,39 62,89±1,82 65,67±1,85
neutrophils, %
Lymphocytes,
22,92±1,19 21,43±1,60 25,25±1,67 22,89±1,89
%
Monocytes, % 3,50±0,33 3,77±0,44 3,07±0,49 3,56±0,74
ESR, mm/hour 18,17±2,06 16,75±2,68 20,39±3,22 16,22±3,51
Glucose, mmol/l 6,23 ± 0,14 5,78 ± 0,10 6,94± 0,18** 5,59± 0,35*
Creatinine, μmol/l 75,46±2,33 70,44±2,64 83,34±3,93** 79,28±3,93
Urea, mmol/l 5,99±0,24 5,65±0,28 6,54±0,43 6,61±0,67
AST, u/l 20,34±1,85 21,29±2,83 18,86±1,73 25,18±6,95
ALT, u/l 20,14±1,34 21,15±1,85 18,56±1,83 24,59±4,30
Bilirubin, 9,81±0,59 9,13±0,70 10,89±1,04 11,31±1,71
μmol/l
К, mmol/l 5,79±1,10 6,54±1,80 4,61±0,14** 5,00±0,25
Na, mmol/l 138,14±0,41 138,91±0,55 136,93±0,53** 138,22±0,77
LDL, mmol/l 3,26±0,14 3,47±0,18 2,92±0,21 3,42±0,28
HDL, mmol/l 1,26±0,05 1,20±0,06 1,35±0,08 1,30±0,11
Cholesterol,
4,56±0,14 4,64±0,20 4,43±0,20 4,74±0,28
mmol/l
Fibrinogen, g/l 4,56±0,18 4,43±0,22 4,77±0,34 4,59±0,39
Prothrombin time,
11,65±0,15 11,81±0,22 11,40±0,18 11,43±0,22
sec./
Prothrombin
according to Kwik, 96,58±2,15 94,31±2,86 100,17±3,16 99,31±3,88
%
INR, index 0,99±0,02 0,99±0,03 0,99±0,02 0,96±0,02
Trombin time, sec./ 11,11±0,14 11,28±0,19 10,83±0,21 10,86±0,37
Remark 1. * – p˂0.05 compared to the 1st group.
Remark 2. ** – p˂0.005 compared to the 2nd group.
We have also carried out analysis of ultrasound examination indicators of 47 patients (2nd group)
who underwent open and endovascular surgical interventions. All the patients were diagnosed with an
insignificant stenosis at the level of the aorta/iliac segment. However, a significant stenosis/occlusion
at the level of the aorto-iliac segment was not detected. The patency of the femoral segment was
identified in (44.68±7.25)% cases, the patency of the deep femoral artery – in (89.36±4.50)% cases, and
the patency of the popliteal segment – in (68.10±6.80)% cases. We have detected patency at the level
of the tibial arteries at the level of the posterior tibial artery in (53.19±7.29)% of patients, the anterior
tibial artery in (68.09±6.80)% of patients, and the peroneal artery in (80.85 ±5.74) % of patients. The
ankle-brachial index (ABI) was (0.53±0.02) %. The average sPO2 index before surgery was
(83.40±0.81) %, and sPO2 after surgery – (92.21±1.10) %.
3.2. Cluster Analysis
In order to establish combined changes of the parameters study, the most significant for predicting
the risk of complications in the postoperative period, we have performed a neural network clustering of
the indicators of the research. At the same time, the rate of complications (C) in the postoperative period
for each patient was defined: "1" in case of absence of complications, and "2" – in case of presence of
some complications. Neural network clustering of the results of the clinical and anamnestic examination
(Fig. 1) was carried out on the basis of the following indicators: age (1), unhealthy habits (2), body mass
index (3), extracranial arteries injury (4), decompensated diabetes (5), uncompensated diabetes (6),
stroke in anamnesis (7), myocardial infarction in anamnesis (8), pathology of the gastrointestinal tract
(9), respiratory failure (10), cardiovascular diseases (11), oncology in anamnesis ( 12), pulmonary
hypertension (13), reduced ejection fraction (14), mid-range reduced ejection fraction (15), thrombosis
of the reconstruction segment (16), myocardial infarction (17), pseudoaneurysm (18), suppuration of
the prosthesis (19) and C - indicator of complications in the postoperative period (20).
Figure 1 shows the results of the indicators clustering program performance. The 1st cluster includes
20.83% of patients, 2nd – 36.11% patients, and 3rd – 43.06% patients.
Figure 1: The results of clustering of patients’ anamnestic indicators.
According to the research, the patients of the 1st cluster have the highest level of complications in
the postoperative period. With the help of cluster portrait, we found out that this cluster has shown the
highest age indicators (1.6%), lesions of extracranial arteries (3.8%), diabetes in the sub- and
decompensation stages ( 4.4%) and respiratory failure (6.7%), as compared to other clusters. The rate
of stroke in the anamnesis (1.1%) exceeded the similar rate in the 3rd cluster, and the rate of diseases
of the cardiovascular system (0.7%) – exceeded the rate in the 2nd cluster.
Besides, a neural network clustering of the results of laboratory-instrumental research (Fig. 2) has
been performed based on a number of indicators: erythrocytes (1), hemoglobin content (2), color index
(3), leukocytes (4), eosinophils (5), rod-shaped neutrophils ( 6), segmented neutrophils (7), lymphocytes
(8), monocytes (9), ESR (10), glucose (11), creatinine (12), urea (13), AST (14), ALT (15), bilirubin
(16), K (17), Na (18), LDL (19), HDL (20), cholesterol (21), prothrombin time (22), Prothrombin
according to Kwik (23), INR (24), thrombin time ( 25), fibrinogen (26), ejection fraction (27), allo-graft
(28), deep fundoplasty (29), autovenous graft (30), hybrid surgery (31), stenting (32), balloon
angioplasty (33), reconstruction segment thrombosis (34) ), myocardial infarction (35),
pseudoaneurysm (36), suppuration of the prosthesis (37), and C – is an indicator of complications in the
postoperative period (38).
Figure 2: Results of indicators clustering of laboratory-instrumental examination of patients
Figure 2 shows the results of clustering of indicators program performance. 1st cluster includes
51.39% of patients, 2nd – 20.83%, and 3rd cluster – 27.78%.
The highest value of the complications indicator in the postoperative period was found out in the 2nd
cluster. With the help of a cluster portrait, we have come to conclusion that the 2nd cluster includes the
highest number of erythrocytes (5.4%), monocytes (10.5%), creatinine levels (5.4%), AST (36.1%),
ALT (34.7%), bilirubin (26.6%), potassium (2.7%), low-density lipoproteins (6.6%) and cholesterol
(5.34%). Indicators of urea (2.2%) and prothrombin according to Kwik (3.1%) exceeded those in the
1st cluster.
We have also performed a neural network clustering of the ultrasound study results (Fig. 3) based
on the following indicators: ultrasound of the aorto-iliac segment (1), ultrasound of the femoral-popliteal
segment (2), hemodynamically insignificant stenosis at the level of the aorto-iliac segment (3) ,
hemodynamically significant stenosis/occlusion at the level of the aorto-iliac segment (4), patency of
the femoral segment (5), patency of the deep femoral artery (6), patency of the popliteal segment (7),
patency of the posterior tibial artery (8), patency of the anterior tibial artery (9), patency of the peroneal
artery (10), ankle-brachial index (11), sPO2 before surgery (12), sPO2 after surgery (13), level of
revascularization (14), thrombosis of the reconstruction segment (15), myocardial infarction (16) ),
embolism (17), pseudoaneurysm (18), suppuration of the prosthesis (19) and C is an indicator of
complications in the postoperative period (20).
As indicated in Figure 3, the highest indicator value of complications in the postoperative period
was found in the 3rd cluster. With the help of a cluster portrait, it can be determined that this cluster
also had the lowest values of femoral segment patency (-16.1%), peroneal artery patency (-1.3%), as
well as the ankle-brachial index (-2.5%). The posterior tibial artery patency value (-2.1%) and anterior
tibial artery patency value (-2.3%) in the 3rd cluster was lower compared to the 1st cluster.
Figure 3: Results of clustering of ultrasound examination indicators.
3.3. Risk scale
Thus, based on the results of neural network clustering, we have identified groups of anamnestic,
laboratory, and ultrasound examination indicators, combined changes of which are the most significant
for predicting the risk of complications in the postoperative period. The obtained results of neural
network clustering have been included into the NeuroXL Classifier program in order to create a scale
for determining the risk of postoperative complications. At the same time, the limit values of their
indicators are determine on the basis of values defined by patients clustering.
The coefficients value for the indicators was set as the ratio of their fractions in the specified cluster
to the minimum fraction of the indicator. Its coefficient was defined as 1.0. Thus, in the group of
anamnestic indicators, the indicator of cardiovascular system diseases had the smallest fraction (0.70%)
in the cluster with the largest number of complications. This indicator is considered to be a unit.
Accordingly, the next-highest fraction of stroke (1.07%) exceeded the previous one by 1.5, so its
coefficient was 1.5. Other coefficients of anamnestic indicators, as well as laboratory indicators and
ultrasound examination indicators, which are defined on the basis of clustering as the most important
factors for predicting, were determined in a similar way.
It should be noted that by adding to the scale indicators, that are not allocated to the most important
groups for prediction based on clustering, but are the risk factors for the development of complications
according to the results of other studies, a minimum coefficient of 1.0 was defined.
To unify the definition of risk levels, all coefficient values were converted into 10-point scales
according to the research directions with formation of scales for anamnestic (Table 2), laboratory (Table
3) indicators, indicators of symptomatic ultrasound examination (Table 4) and contralateral ultrasound
examination (Table 5) of patients’ limbs. The maximum number was 40 points, respectively.
Indicators of all clustering analysis, such as anamnestic, laboratory, ultrasound examination of the
symptomatic and contralateral limbs indicators, have been included into the NeuroXL Classifier
program in order to define the values of their points. Names of some indicators, including ultrasound,
have been adapted to facilitate the use in the vascular surgery department.
Table 2
Values of coefficients and scores for anamnestic indicators of patients with open and endovascular
surgical interventions.
Indicator Coefficient Score
Age ≥ 65 years 2,3 0,7
Bad Habits 1,0 0,3
Body mass index ≥ 22,6 1,0 0,3
Atherosclerotic carotid disease 5,4 1,6
Diabetes (in the stage of 1,0 0,3
compensation)
Diabetes (in the stage of sub- and 6,2 1,8
decompensation)
History of stroke 1,5 0,4
History of myocardial infarction 1,0 0,3
Respiratory failure 9,5 2,8
Pulmonary hypertension 1,0 0,3
Diseases of the cardiovascular system 1,0 0,3
Heart failure with reduced left 1,0 0,3
ventricular ejection fraction ≤ 49%
Pathology of the gastrointestinal tract 1,0 0,3
History of oncological diseases 1,0 0,3
Table 3
Values of coefficients and scores for laboratory indicators of patients with open and endovascular
surgical interventions
Indicator Value Coefficient Score
Erythrocytes ≥4,5 2,5 0,4
Monocytes ≥3,6 4,8 0,8
Creatinine ≥79,3 2,5 0,4
Urea ≥6,6 1,0 0,1
AST ≥25,2 16,4 2,6
ALT ≥24,6 15,8 2,5
Bilirubin ≥11,3 12,1 1,9
К ≥5,0 1,2 0,2
LDL ≥3,4 3,0 0,5
Cholesterol ≥4,7 2,4 0,4
Thrombin time ≥99,3 1,4 0,2
Table 4
Values of coefficients and scores for indicators of ultrasound examination of the symptomatic limbs of
patients with open and endovascular surgical interventions.
Indicator Coefficient Score
Stenosis in the range of 51-70% at the level
1,0 0,4
of the aorto/iliac segment
Stenosis ≥ 71% or occlusion at the level of
1,0 0,4
the aorto-iliac segment
Stenosis ≥ 71% or occlusion at the level of 12,4 5,3
the femoral segment
Stenosis ≥ 71% or occlusion at the level of 1,0 0,4
the deep femoral artery
Stenosis ≥ 71% or occlusion at the level of 1,0 0,4
the a.poplitea
Stenosis/occlusion of a. tibialis posterior 1,6 0,7
Stenosis/occlusion of a. tibialis anterior 1,8 0,8
Stenosis/occlusion of peroneal artery 1,0 0,4
Ankle-brachial index ≤ 0,53 1,9 0,8
sPO2 before surgery ≤ 83,4 1,0 0,4
Table 5
Values of coefficients and points for ultrasound examination indicators of the contra-lateral limbs of
patients with open and endovascular surgical interventions
Indicator Coefficient Score
Stenosis in the range of 51-70% at the level
1,0 0,4
of the aorto/iliac segment
Stenosis ≥ 71% or occlusion at the level of
1,0 0,4
the aorto-iliac segment
Stenosis ≥ 71% or occlusion at the level of 12,4 5,3
the femoral segment
Stenosis ≥ 71% or occlusion at the level of 1,0 0,4
the deep femoral artery
Stenosis ≥ 71% or occlusion at the level of 1,0 0,4
the a.poplitea
Stenosis/occlusion of a. tibialis posterior 1,6 0,7
Stenosis/occlusion of a. tibialis anterior 1,8 0,8
Stenosis/occlusion of peroneal artery 1,0 0,4
Ankle-brachial index ≤ 0,53 1,9 0,8
sPO2 before surgery ≤ 83,4 1,0 0,4
Table 6
Cumulative scoring system for assessing the risk of developing complications.
Anamnestic parameters
Indicator Score
Age ≥ 65 years 0,7
Bad Habits 0,3
Body mass index ≥ 22,6 0,3
Atherosclerotic carotid disease 1,6
Diabetes (in the stage of compensation) 0,3
Diabetes (in the stage of sub- and decompensation) 1,8
History of stroke 0,4
History of myocardial infarction 0,3
Respiratory failure 2,8
Pulmonary hypertension 0,3
Diseases of the cardiovascular system 0,3
Heart failure with reduced left ventricular ejection fraction 0,3
≤ 49%
Pathology of the gastrointestinal tract 0,3
History of oncological diseases 0,3
Laboratory indicators
Indicator Score
Erythrocytes ≥ 4,5 *1012 /l 0,4
Monocytes ≥ 3,6 % 0,8
Creatinine ≥ 79,3 μmol/l 0,4
Urea ≥ 6,6 mmol/l 0,1
AST ≥ 25,2 u/l 2,6
ALT ≥ 24,6 u/l 2,5
Bilirubin ≥ 11,3 μmol/l 1,9
К ≥ 5,0 mmol/l 0,2
LDL ≥ 3,4 mmol/l 0,5
Cholesterol ≥ 4,7 mmol/l 0,4
Thrombin time ≥ 99,3 % 0,2
Ultrasound examination indicators of symptomatic limb
Indicator Score
Stenosis in the range of 51-70% at the level of the
0,4
aorto/iliac segment
Stenosis ≥ 71% or occlusion at the level of the aorto-iliac
0,4
segment
Stenosis ≥ 71% or occlusion at the level of the femoral 5,3
segment
Stenosis ≥ 71% or occlusion at the level of the deep femoral 0,4
artery
Stenosis ≥ 71% or occlusion at the level of the a.poplitea 0,4
Stenosis/occlusion of a. tibialis posterior 0,7
Stenosis/occlusion of a. tibialis anterior 0,8
Stenosis/occlusion of peroneal artery 0,4
Ankle-brachial index ≤ 0,53 0,8
sPO2 before surgery ≤ 83,4 0,4
Ultrasound examination indicators of contralateral limb
Indicator Score
Stenosis in the range of 51-70% at the level of the
0,4
aorto/iliac segment
Stenosis ≥ 71% or occlusion at the level of the aorto-iliac
0,4
segment
Stenosis ≥ 71% or occlusion at the level of the femoral 5,3
segment
Stenosis ≥ 71% or occlusion at the level of the deep femoral 0,4
artery
Stenosis ≥ 71% or occlusion at the level of the a.poplitea 0,4
Stenosis/occlusion of a. tibialis posterior 0,7
Stenosis/occlusion of a. tibialis anterior 0,8
Stenosis/occlusion of peroneal artery 0,4
Ankle-brachial index ≤ 0,53 0,8
sPO2 before surgery ≤ 83,4 0,4
According to the obtained results of the scoring system for defining the risk of postoperative
complications, we have developed a scale of the level of risk of complications in the postoperative
period of patients with diseases of the main arteries that underwent open and endovascular surgical
interventions (Table 7). Accordingly, we have defied the risk of the development of postoperative
complications, in the case of aggregate points value for all directions of our research: a very high risk -
31-40, a high risk - 21-30, a medium risk - 11-20, and a low risk - 1-10.
Table 7
Cumulative scoring system for assessing the risk of developing complications.
Total points in all areas of research
Risk rate Total points
Low risk 1–10
Moderate risk 11–20
High risk 21–30
Very high risk 31–40
The following is a suggested range of scales for defining postoperative complications: SVS Wifi
(2019), Finnvasc (2007), Prevent III (2006), BASIL (2010), ERICVA (2016), Caprini (1991), GCS
(1974) [16]. Each of them takes into account separate criteria of the pathology of organs and systems
and their influence on the main pathology.
The suggested scale of risk stratification of the development of complications of operative treatment
of main arteries of the lower extremities in the postoperative period takes into account the multifactorial
nature of clinical, anamnestic and laboratory-instrumental studies. Consideration of the combination of
factors that characterize the state of organs and systems affects the choice and method of reconstructive
surgery.
4. Conclusions
In order to predict possible postoperative complications of surgical interventions into the great
vessels of the lower limbs, we have defined a level of risk of developing complications by processing
them by neural network clustering of indicators of clinical, anamnestic and laboratory-instrumental
studies, followed by the NeuroXL Classifier processing program.
Based on the determined level of possible postoperative complications of surgery on the main arteries
of the lower limbs, we have defined four levels of risk of developing complications: 31-40 – a very high
risk, 21-30 – a high risk, 11-20 – a moderate risk, and 1-10 – a low risk.
Acknowledgements
There were no external sources of funding and support. No fees or other compensation were paid.
The authors who participated in this study declared that they have no conflict of interest regarding this
manuscript.
5. References
[1] Bishop, C. (1996) “Neural networks,” Handbook of Neural Computation [Preprint]. Available at:
https://doi.org/10.1201/9781420050646.ptb6.
[2] Du K. Clustering: A neural network approach. Neural Networks. 2010;23(1):89-107. doi:
10.1016/j.neunet.2009.08.007.
[3] Martsenyuk V, Selskyi P, Tvorko V. Analiz rezultativ obstezhennia patsiientiv z hipertenziieiu na
osnovi koreliatsiinykh pokaznykiv ta bahatoparametrychnoi neiromerezhevoi klasteryzatsii z
metoiu optymizatsii prohnozuvannia perebihu zakhvoriuvannia na pervynnomu rivni. Medical
Informatics and Engineering. 2013;(2). doi: 10.11603/mie.1996-1960.2013.2.1720. [In Ukrainian].
[4] Selskyy P, Vakulenko D, Televiak A, Veresiuk T. On an algorithm for decision-making for the
optimization of disease prediction at the primary health care level using neural network clustering.
Family Medicine & Primary Care Review. 2018;20(2):171-175. doi: 10.5114/fmpcr.2018.76463.
[5] Zhao Z, Voros S, Weng Y, Chang F, Li R. Tracking-by-detection of surgical instruments in
minimally invasive surgery via the convolutional neural network deep learning-based method.
Computer Assisted Surgery. 2017;22(sup1):26-35. doi: 10.1080/24699322.2017.1378777.
[6] Filiberto A, Loftus T, Elder C, Hensley S, Frantz A, Efron P et al. Intraoperative hypotension and
complications after vascular surgery: A scoping review. Surgery. 2021;170(1):311-317. doi:
10.1016/j.surg.2021.03.054.
[7] Harris D, Herrera A, Drucker C, Kalsi R, Menon N, Toursavadkohi S et al. Defining the burden,
scope, and future of vascular acute care surgery. Journal of Vascular Surgery. 2017;66(5):1511-
1517. doi:10.1016/j.jvs.2017.04.060.
[8] Dominioni L, Imperatori A, Rotolo N, Rovera F. Risk Factors for Surgical Infections. Surgical
Infections. 2006;7(supplement 2):s-9-s-12. doi.10.1089/sur.2006.7.s2-9.
[9] Hentati H, Lim C, Salloum C, Azoulay D. Authors’ Reply: Risk Factors for Mortality and
Morbidity in Elderly Patients Presenting with Digestive Surgical Emergencies. World Journal of
Surgery. 2018;42(12):4129-4129. doi: 10.1007/s00268-018-4701-z.
[10] Sobczak F, Pais-Roldán P, Takahashi K, Yu X. Decoding the brain state-dependent relationship
between pupil dynamics and resting state fMRI signal fluctuation. eLife. 2021;10. doi:
10.7554/elife.68980.
[11] Fritz B, Marbach G, Civardi F, Fucentese S, Pfirrmann C. Deep convolutional neural network-
based detection of meniscus tears: comparison with radiologists and surgery as standard of
reference. Skeletal Radiology. 2020;49(8):1207-1217. doi: 10.1007/s00256-020-03410-2.
[12] Kobza I, Yarema Y, Zhuk R, Fedoriv D. Rekonstruktyvni operatsii na arteriiakh stopy v likuvanni
krytychnoi ishemii nyzhnikh kintsivok.. UMJ Heart & Vessels. 2018;0(1):37-39. doi:
10.30978/hv2018137. [In Ukrainian].
[13] Hicks, C., Najafian, A., Farber, A., Menard, M., Malas, M., Black, J. and Abularrage, C., 2016.
Below-knee endovascular interventions have better outcomes compared to open bypass for patients
with critical limb ischemia. Vascular Medicine, 22(1), pp.28-34. doi: 10.1177/1358863x16676901.
[14] Gentile, F., Lundberg, G. and Hultgren, R., 2016. Outcome for Endovascular and Open Procedures
in Infrapopliteal Lesions for Critical Limb Ischemia: Registry Based Single Center Study.
European Journal of Vascular and Endovascular Surgery, 52(5), pp.643-649. doi:
10.1016/j.ejvs.2016.07.013.
[15] Matsagkas, M., Kouvelos, G., Arnaoutoglou, E., Papa, N., Labropoulos, N. and Tassiopoulos, A.,
2011. Hybrid Procedures for Patients With Critical Limb Ischemia and Severe Common Femoral
Artery Atherosclerosis. Annals of Vascular Surgery, 25(8), pp.1063-1069. doi:
10.1016/j.avsg.2011.07.010.
[16] Abualhin, M., Gargiulo, M., Bianchini Massoni, C., Mauro, R., Morselli-Labate, A., Freyrie, A.,
Faggioli, G. and Stella, A., 2019. A prognostic score for clinical success after revascularization of
critical limb ischemia in hemodialysis patients. Journal of Vascular Surgery, 70(3), pp.901-912.
doi: 10.1016/j.jvs.2018.11.034.
[17] Venher, I., Kostiv, S., Selskyi, B., Faryna, I., Orlov, M., Tsiupryk, N., Kovalskiy., 2022.
Intraoperative levels of coagulation factors in patients treated with open and endovascular
revascularization of occluded tibial arteries. Georgian Med News, 2(323), pp.11-17. pmid:
35271465.
[18] Venher, I., Rusin, V., Kostiv, S., Zarudna, O., Kostiv, O.I., 2017. Hypercoagulable Syndrome in
the Early Postoperative Period is a Factor of Venous Thromboembolism. Novosti Khirurgii. 3(25),
pp. 267-272. doi: 10.18484/2305-0047.2017.3.267.