You are on page 1of 17

Chaos, Solitons and Fractals 153 (2021) 111445

Contents lists available at ScienceDirect

Chaos, Solitons and Fractals


Nonlinear Science, and Nonequilibrium and Complex Phenomena
journal homepage: www.elsevier.com/locate/chaos

Review

Based on the multi-assessment model: Towards a new context of


combining the artificial neural network and structural equation
modelling: A review
A.S. Albahri a,d, Alhamzah Alnoor c, A.A. Zaidan b, O.S. Albahri b, Hamsa Hameed a,∗,
B.B. Zaidan e, S.S. Peh a, A.B. Zain a, S.B. Siraj a, A.H. Alamoodi b, A.A. Yass f
a
Faculty of Human Development, Sultan Idris University of Education (UPSI), Tanjung Malim, Malaysia
b
Department of Computing, Sultan Idris University of Education (UPSI), Tanjong Malim, Malaysia
c
School of Management, Universiti Sains Malaysia, 11800 Pulau Pinang, Malaysia
d
Informatics Institute for Postgraduate, Studies (IIPS), Iraqi Commission for Computers and Informatics, Baghdad, Iraq
e
National Yunlin University of Science and Technology, Future Technology Research Center, Yunlin, Taiwan, ROC
f
IT Infra Expert Company, Earthlink, Main ISP, Baghdad, Iraq

a r t i c l e i n f o a b s t r a c t

Article history: Theoretical models have become increasingly complex, but the dual-phase structural equation modelling
Received 2 May 2021 (SEM) and artificial neural network analysis can be used by scholars to unveil the causal interactions
Revised 4 July 2021
and nonlinear relationships between variables. However, not only a single open issue and challenge—but
Accepted 13 September 2021
several of them—are encountered in the use of different multi-assessment types of measurement model
Available online 14 November 2021
to achieve the reliability and validity whilst implementing SEM, but the gaps have not been fully deter-
Keywords: mined at present. The issues significantly impact the effectiveness process of selecting the most suitable
Structural equation modelling method to assess the measurement model of SEM. Once the best sequence quality improvement is met, it
Partial least square−structural equation then needs to present a recommendable solution. To this end, this study completes the literature by pre-
modelling senting a systematic review of all main advanced aspects of the SEM reliability and validity approaches.
Artificial neural network Firstly, the databases of ScienceDirect, IEEE Xplore, Web of Science and Scopus were checked for the
Multi-criteria decision making
retrospective studies. A total of 239 papers were gathered for the period covering 2016 to June 2021.
Then, the obtained articles were filtered according to the predefined inclusion criteria. Sixty articles were
ultimately selected and divided into three categories (single, hybrid and other types) to enable a new
representation of the crossover taxonomy amongst ‘SEM reliability and validity’ and ‘multi-assessment
methods for structural model’ for the first time. The three categories had been matched with the SEM
processes, and each of the detailed models were defined to determine the sets of principal criteria of the
entire selected SEM approaches. Consequently, this multi-field interdisciplinary review was used to ex-
pose the state-of-the-art challenges and open issues (i.e. multiple-evaluation criteria, importance criteria
and data variation) related to the sets of SEM criteria necessitating a selection process for deriving the
best SEM method. Each issue entailed a ‘wherefore’, and multi-criteria decision making was adopted to
handle the complexity problems in the different cases. Thus, a new three-phase decision-making method-
ology was constructed. In the first phase, a decision matrix (DM) was identified for the SEM approach;
the composition of the decision alternatives and identified criteria were derived from the academic lit-
erature. In the second phase, the development methodology was achieved on the basis of the integrated
multi-criteria DM techniques. The analytic hierarchy process was used for the subjective weighting of
the criteria within the constructed DM, whereas the vlsekriterijumska optimizcija i kaompromisno resenje


Corresponding author.
E-mail address: hamsahameed@fpm.upsi.edu.my (H. Hameed).

https://doi.org/10.1016/j.chaos.2021.111445
0960-0779/© 2021 Elsevier Ltd. All rights reserved.
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

technique was used for ranking and selecting the best SEM methods. In the third phase, an objective vali-
dation approach was adopted to validate the proposed methodology. The outcome of this novel approach
is intended to guide decision makers and policymakers on the easy evaluation of their goals of selecting
the most suitable computing methods and the improvement of the reliability and validity of SEM.
© 2021 Elsevier Ltd. All rights reserved.

a significant concern to the accurate input to ANN [3,22,28,44,49–


1. Introduction 53]. The present study can attain the goal of answering the above-
mentioned question and contribute to the body of PLS-SEM knowl-
Structural equation modelling (SEM) is a popular method for edge by presenting recent studies and conducting a systematic
parameter estimation and hypothesis testing [1–6]. The SEM tech- literature review (refer to the ‘Analysis of Multi-Assessment SEM
nique contributes to the development of exploratory research the- model’ section). The method, information source and study selec-
ories by explaining the variance that contributes to an estimate tion, amongst others, are identified and explained to establish the
of linear equations, which include both observed and latent vari- crossover taxonomy amongst the ‘SEM reliability and validity’ and
ables [7–9]. SEM enables researchers to simultaneously examine assessment methods for the structural model (Fig. 2).
measurement and theoretical concepts [10–15]. Furthermore, SEM The systematic literature review provides exclusive evidence
has become a mainstream tool in social research and is employed showing that the previous studies entail different types of multi-
continuously. Authors from numerous nations have contributed assessment of measurement methods, which can substantially im-
to the implementation of numerous second-generation technolo- pact the inputs of upcoming processes of SEM and other related
gies (e.g. partial least square−structural equation modelling [PLS- processes, such as ANN and contribution processes [22]. However,
SEM]) [16–19]. This method is useful for developing and access- the three specific challenges with respect to SEM that can influ-
ing theories and assessing the interaction of factors, in which the ence the explicit relations between explanatory variables (input)
conceptual constructs include independent, mediating, moderat- and dependent variables (output) had not been considered [54,55].
ing and dependent factors via multi-assessment of measurement Therefore, what are the main challenges and issues related to the im-
modelling (i.e. reliability and validity representing an outer model) plementation of SEM calculation? This discussion leads to a genuine
and structural model assessment (i.e. hypotheses tests representing concern regarding effectiveness, namely, the effect of the SEM out-
an inner model) [20–22]. Hence, the PLS-SEM method is appro- put when different multi-assessment methods lead to varying re-
priate for explaining causal relationships; it has predictive advan- sults or effects on ANN and contribution process [22,34,50]. Be-
tages, such as the use of the coefficient determination (R2 ), pre- sides, the methods of establishing the reliability and validity in
dictive relevance (Q2 ) and effect size (F2 ) values [23–26]. How- SEM is a concern related to the overall SEM output results based
ever, the PLS-SEM method cannot explore the nonlinear relation- on the R2 , F2 and Q2 indicators that consider the significant criteria
ship amongst constructs [27–31]. Therefore, the authors had be- of the SEM model (refer to the ‘Open Issues and Challenges’ sec-
come interested in solving this problem by attempting to improve tion). In these contexts, the aforementioned issue raises another
the performance of the PLS-SEM technique for exploring the non- important question: What is the recommended solution for such
linear and non-compensatory relationships between constraints, in challenging aspects and their issues when selecting the best method
which the quadratic and interaction terms of constraints are inte- as a means of increasing the reliability and validity of SEM mod-
grated into a conceptual framework [32–34]. However, one of the els? Consequently, a novel methodology should be developed to
drawbacks of these methods is the assumption that variables are overcome the SEM challenges regarding effectiveness when select-
normally distributed. When this assumption is breached, the PLS- ing the most suitable method of assessing the SEM measurement
SEM estimates become biased [35–38]. model; this aspect can be answered using new integrated phases
The use of deep learning-based dual-stage PLS-SEM and artifi- (refer to the ‘Methodology’ section).
cial neural network (ANN) analysis provides more in-depth testing
findings than a single-step PLS-SEM method and allows for the ex-
2. Methods
cavation and prediction of variables [39–41]. Hence, ANN analysis
has become a critical method complementing the work of PLS-SEM
This research was developed in accordance with the selected
in recent years [42–44]. The benefits of combining PLS-SEM and
reporting items for meta-analytic and systematic review guide-
ANN are the validation of the PLS-SEM results and the explanation
lines [56,57] (Fig. 1). The essential keyword of this research was
of the nonlinear interactions between antecedents and dependent
‘SEM−ANN’. The scope was limited to the English literature with
variables [45–48].
articles related to SEM−ANN.
As part of our ongoing research intending to clarify how PLS-
SEM and ANN can be integrated with each other, we found an in-
crease in literature that had used PLS-SEM with ANN. However, the 2.1. Information Sources
different study cases that linked PLS-SEM with ANN used varying
types of multi-assessment of measurement models, such as load- According to prior research [32−34], a thorough systematic re-
ing factor, Cronbach’s alpha (CA), average variance extraction (AVE), view should be performed across several databases to include the
composite reliability (CR), Fornell−Larcker, HTMT, Fornell−Larcker majority of publications. Thus, four digital databases were selected
and HTMT, cross-loadings, exploratory factor analysis and confir- for the search of target articles. The selection criteria were de-
matory factor analysis. A complete explanation of how these types signed to offer a much wider perspective on the researchers’ at-
affect PLS-SEM and ANN performance is still incomplete and needs tempt to consolidate a diverse but relevant range of disciplines
to be investigated. Thus, the first question arises: What studies [58–61]. The four databases included the following: 1) IEEE Xplore
had used multi-assessment of the measurement model (outer model) technical literature library on engineering and technology; 2) Sci-
methods to examine reliability and validity? The literature differed in enceDirect database, which offers access to technology and science
calculating the reliability and validity of the various methods and journal articles; 3) Web of Science (WoS), which involves cross-
obtained varying values of R2 between PLS-SEM and ANN, posing disciplinary research materials; and 4) Scopus library of cross-

2
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

Fig. 1. Systematic review protocol.

disciplinary literature. These databases were selected for their aca- 2.4. Eligibility criteria
demic credibility and breadth of coverage across academic fields.
The eligibility criteria included the following:

2.2. Study selection 1. The papers were published in a journal or as a conference pa-
per in the English language.
The research selection process began with a search of the avail- 2. Attention was given on SEM and ANN analysis.
able literature, followed by two rounds of screening and filtering
All articles that conformed with the above criteria were se-
[62–65]. The first iteration eliminated duplicate and irrelevant ar-
lected. In summary, the initial purpose was to map the SEM and
ticles based on scanned titles and abstracts [66,67]. After a com-
ANN model cases and concerns into a general and coarse-grained
prehensive full-text reading of the articles that were screened in
review. The inclusion/exclusion criteria included studies not related
the first phase, the articles were filtered in the second iteration
to the SEM and ANN models or other irrelevant topics.
[68,69]. The same qualifying criteria were used in both versions.

3. Analytic results of the multi-assessment SEM model


2.3. Search protocol
In testing the hypotheses in SEM, the multi-assessment of the
A search protocol was performed at the beginning of June 2021 measurement model (outer model) should be considered before
by using the abovementioned databases. The Boolean query search assessing the structural model (inner model). Several ways can be
utilised numerous keywords correlated with ‘structural equation used to verify this aspect and determine the factors that charac-
modeling’ and ‘artificial neural network’. The keywords in the terise reliability and validity. As mentioned in Section 1, the four
search process contained a combination of words and operands methods for assessing the measurement model can also be in-
(Fig. 1). cluded into the sub-methods. Consequently, only the publications

3
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

that could meet the criteria outlined in Fig. 1 were included in our
study. We obtained 239 entries from the selected databases. After

Exploratory Factor Analysis Confirmatory Factor Analysis


eliminating the 12 duplicate records, 227 articles were retained.
Following an examination of the titles and abstracts, 84 of them
fulfilled the original selection criterion. Then, after examining the
complete texts of the 60 publications, all of them were found to
fulfil the inclusion criteria and thus included in the final evalua-
tion. The final evaluation was split into three main groups. The first
category, which contained 7 articles (12%), represented the ‘sin-
gle based method’. In this category, a unique method of measuring


the reliability and validity (i.e. loading factor, CA, AVE and CR) was
conducted for the articles. Loading factor particularly refers to the
load factors for items exceeding 0.7. CA measures reliability [70].
AVE evaluates the degree of variation that is recorded by a con-
struct versus the level attributable to the measurement error; val-
ues higher than 0.7 are regarded as excellent, whereas levels less
than 0.5 are regarded as acceptable [71]. CR is a less biased assess-

11%
ment of reliability than the CA; an appropriate CR value is 0.7 or


higher.

Fornell–Larcker HTMT Fornell–Larcker and HTMT Cross loadings


The second category, which contained 44 articles (73%), repre-
sented the ‘hybrid based method’. This category included articles
that had used three types of discriminant validity methods, in-

3%
cluding the convergent validity method. For this category, firstly,


the Fornell–Larcker testing technique measures discriminant valid-
ity by comparing the amount of variance captured by the con-
struct (i.e. AVE) and the shared variance with other constructs
[72]. Secondly, the heterotrait−monotrait (HTMT) ratio approach
for correlations assesses the discriminant validity [73]. Thirdly, the
cross-loadings may be linked back to exploratory factor analysis, in

Multi-Assessment of the Measurement Model (Outer Model)


which researchers examine the indicator loading patterns to iden-
tify those indicators with high loadings on a single factor and the

3%

indications with high loadings on several factors [72].

11%
The third category, which contained 9 articles (15%), repre- Discriminant validity


sented ‘others’. This category involved articles depicting other
methods of measuring reliability and validity. The first method was
exploratory factor analysis, which can explore the basic dimensions
of variables. The second method was confirmatory factor analysis,
which can be used for testing the hypotheses to determine the best

83% 83% 51% 6%


model [74].
AVE CR

The academic studies in the literature entailed different sub-





methods to achieve reliability and validity. Therefore, we explored


Convergent validity




the SEM processes as a means of offering a comprehensive analysis


of the collected studies on the cited methods within the scope of
CA




the aforementioned three categories, subsequently completing our


Loading

argument. The details are listed in Table 1.


83%

The loading, CA, CR, AVE and discriminant validity methods had





been applied most frequently and were distributed amongst the


[4,8-13,16,18,23-27,32-34,36,41,45-47,80-90]

single and hybrid categories. The application of exploratory factor


analysis and confirmatory factor analysis were limited in the stud-
ies belonging to the ‘others’ category. The mean value is computed
as follows:
total f requence
mean = × 100. (1)
N
[3,22,28,44,49,50,91]

[7,17,19,27,29,95,96]

In addition, although Ref. [97] stressed the need to use both


convergent validity and discriminant validity when applying SEM
[5,14,43,75-77]

to the multi-assessment of measurement model, many problems


[6,40,92,93]

and issues were observed regarding this topic. Moreover, discrep-


[42,94]
[79,80]

ancies were apparent amongst the studies in terms of the use of


[78]

85%
Category Type Ref.

appropriate methods for measuring reliability and validity. The per-


centage of the reviewed papers ranged between 3% to 85% for the
SEM processes.

multi-assessment of the measurement model for determining the


factors that can be subject to hypothesis testing and ANN input ob-
Hybrid

Others
Single

Mean
Table 1

tainment. A significant concern is thus raised with regards to the


adoption of any method to integrate SEM with ANN. The differ-
ences in the methods of reliability and validity measurement are

4
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

Fig. 2. Crossover taxonomy amongst ‘SEM reliability and validity’ and multi-assessment methods for the structural model.

explained in Ref. [98], and it also confirms the variation in the re-
sults of these methods.
Furthermore, the trends in the literature review seem to indi-
cate a consolidated and improved approach for the main categories
involving the crossover in the utilisation of the abovementioned
multi-assessment measuring modelling techniques. The crossover
taxonomy is shown in Fig. 2.
The crossover taxonomy can help to explain the matching be-
tween each category and the multi-assessment methods for the
structure model. The hybrid category shows four directions for
the integrated multi-assessment methods for ‘loading’, ‘CA’, ‘AVE’
and ‘CR’ along with ‘Fornell–Larcker’, ‘HTMT’, ‘Fornell–Larcker and
HTMT’ and ‘cross-loading’. However, no integration direction to-
wards ‘exploratory factor analysis’ and ‘confirmatory factor anal-
ysis’ is apparent. Thus, the literature varies in terms of how to use
the methods. This scenario signifies a need for a novel solution to Fig. 3. Number of articles by category and publication journal.
select the best method and/or direction to be able to measure re-
liability and validity in SEM (refer to Section 9).
rect, 31 papers; WoS database, 24 papers; IEEE Xplore, 1 paper;
and Scopus, 4 papers.
The SEM–ANN papers in ScienceDirect represent the highest
4. Comprehensive science mapping analysis percentage of total articles. This high number can be attributed
to the authors relating many of ScienceDirect’s journals to the so-
Comprehensive science mapping analysis can be adopted to ex- cial sciences. The number of articles published in WoS and Sco-
plore the problems related to the SEM–ANN integration, as most pus may have decreased due to their limited number of journals
cases in the literature largely focus on the hybrid method’s ad- in the social science field. The literature dealing with hybrid va-
vantages. Moreover, as many barriers had been neglected in the lidity methods (i.e. convergent validity and discriminant validity),
literature, a mapping analysis based on the bibliometrix is neces- such as those published in ScienceDirect’s and WoS’ journals, has
sitated [99]. Four digital databases have stored numerous types of increased. Most of the articles focus on convergent validity analy-
research publications, as shown in Fig. 3. The total numbers of the sis by using the abovementioned four indicators and discriminant
published articles selected in this study are as follows: ScienceDi- validity analysis by using the Fornell–Larcker method. However,

5
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

Fig. 4. Annual scientific production.

the Fornell–Larcker criterion reportedly does not perform well [72]. broad prospect for various future studies, and they may even open
Thus, the multi-assessment SEM model should be investigated be- new horizons of cooperation in the SEM–ANN field.
cause the aforementioned issues may arise when the databases are A research cooperation network is a type of network in which
eventually utilised. the authors are the nodes and the co-authors are the contacts. Re-
search cooperation is one of the most recorded ways of teamwork
4.1. Annual scientific production in science. Hence, the strength of the SEM–ANN research network
may also represent the existence of a network of cooperation be-
Fig. 4 shows the relationship between annual scientific produc- tween or amongst authors.
tion and year of publication. The trend depicts a dramatic increase
in interest with June 2021. 5. Multi-assessment model in SEM combined with ANN
The annual scientific production in 2021 represents the largest
number of papers for the study period. Most of them refer to the This section presents the state-of-the-art processes of the
hybrid-based method, followed by the exploratory factor analy- multi-assessment model in SEM combined with the ANN workflow
sis and confirmatory factor analysis. The rise in articles about the and its complex relationships (Fig. 7).
hybrid-based method may be attributed to the numerous problems The implementation of the SEM process begins with the in-
associated with this topic in recent years. The combination of SEM ner and outer models (i.e. multi-assessment of measurements and
and ANN analysis is vital for discovering the causal and nonlinear assessment of the structural model), as they represent the out-
relationships amongst the constructs. This advantage can explain puts of SEM. These inner and outer models are the same as the
why the studies focusing on this type of analysis have increased in feeder inputs to the ANN and contribution processes. However,
recent years. In particular, the increase is 37% in 2020, followed by in SEM, the multi-assessment of measurements entails the fol-
17% in 2021. The rise may also be explained by the usefulness of lowing sub-methods: convergent validity, discriminant validity, ex-
ANN, as this approach provides the ranks and validates both sig- ploratory factor analysis and confirmatory factor analysis. In con-
nificant predictors and the SEM results. vergent validity, four types of multi-assessment measurements (i.e.
loading, CA, AVE and CR) are mentioned in the literature [97]. In
4.2. Wordcloud their specific contexts, the loading factor should exceed 0.7, AVE
should exceed 0.5 and CR and CA (for internal consistency relia-
Wordcloud is a tool used to identify the most important top- bility) should exceed 0.7. Discriminant validity includes the follow-
ics dealing with a certain subject matter. Fig. 5 shows the essen- ing three methods: Fornell–Larcker, HTMT and cross-loading. How-
tial critical words adopted by the previous studies about SEM–ANN ever, in the literature, other methods, particularly exploratory fac-
and the relationships amongst them. tor analysis or confirmatory factor analysis, had been used to mea-
Wordcloud can provide support to the research taxonomy by sure reliability and validity. As for SEM, Ref. [98] recommended the
dividing the SEM–ANN studies into three categories. As shown in use of both convergent validity and discriminant validity instead of
Fig. 5, the most frequently used keyword is technology, whether other methods. Then, an assessment of the structural model (inner
they may be about mobile payment or shopping perception, tech- model) is performed. The process involves the testing of hypothe-
nology acceptance and cloud computing. The highest percentage ses whose processes differ depending on the type of variables. The
based on frequent words was for UTAUT at 13%, followed by be- hypotheses may be supported or not. If the hypotheses are sup-
havioural intention at 12%. ported, then the process will move forward to the ANN and con-
tribution processes.
4.3. Country collaboration map
6. Open issues and challenges
Fig. 6 shows a mapping of the scientific cooperation of authors
based overseas. European−African cooperation and European– The questions raised in the previous studies form the open
American cooperation are both lacking. These gaps constitute a challenges and issues. The questions related to the main challenge

6
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

Fig. 5. Most frequent words.

Fig. 6. Country collaboration map.

can be summarised as selection. In detail, for the topic concern- the literature. For example, in Ref. [78], the loading factor was used
ing SEM, the problem of selection is about knowing which method to measure convergent validity, but the measurement of discrimi-
is the best choice for assessing reliability and validity. Fig. 8 and nant validity was neglected. All of the other studies in the litera-
Table 1 show four methods that can be used to assess reliability ture used methods to measure convergent validity. Instead of fo-
and validity. However, the different results depicted by the studies cusing on a single method to measure discriminant validity, an op-
may affect the SEM output results—a challenge that is evident in posing approach was adopted. This finding from the literature re-

7
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

ness (B), the supported hypotheses are affected by the input to


the ANN and contributions to the other process. The effectiveness
concerns consequently raise several practical technical issues, in-
cluding those related to the multiple-evaluation criteria, criteria of
importance and data variation, which are related to the selection
challenge.

6.1. Concerns related to the multiple-evaluation criteria

The multiple-evaluation criteria must be considered when ap-


plying the SEM processes. Three sets of multiple-evaluation cri-
teria can affect the selection process. In the SEM processes, sev-
eral methods of reliability and validity should be considered in the
multi-assessment of the measurement model, as they involve mul-
tiple calculations of the R2 , Q2 and F2 indicators [22,34,50]. In con-
clusion, by taking into account all of the abovementioned differ-
ences regarding each set of criteria, the evaluation phase’s set of
criteria will eventually affect the efficiency of the framework and
the selection process. Thus, the process of selection with respect
to the multiple-evaluation criteria for each of the SEM processes
should be considered. In this regard, a multi-attribute decision ma-
trix (DM) can be used. Fig. 10 shows how the respective set of mul-
tiple criteria for each of the SEM processes can affect the selection
[100].

Fig. 7. Multi-assessment model in SEM combined with ANN.


6.2. Concerns related to the criteria of importance

The importance criteria of the SEM methods are the concern of


this study. Several criteria need to be considered. Different weights
are often given to each set of criteria, either subjectively by deci-
sion makers (experts) [101,102] or objectively via the fixed-weight
method [103], but they further increase the complexity of the task
and thus affect the selection process. The main issue is know-
ing which amongst the criteria is more important than the others.
Moreover, the most appropriate importance criteria in the set can
boost the ranking operation and the selection of the best meth-
ods and runs. Fig. 11 shows how each set of SEM criteria should
be evaluated when detecting the importance of each criterion with
respect to another criterion, a process that can solve the criteria
importance issue.
The criteria weights of the DMs differ in their importance de-
pending on the subjective or objective method of the evaluation.
For example, the SEM criteria (R2 s, Q2 and F2 ) are varied, and the
R2 s may take the higher importance weight in the set [6,40].

Fig. 8. SEM selection process. 6.3. Concerns related to data variation

In this study, data variation widely occurs amongst the dif-


view further confirms the selection problems with respect to SEM ferent SEM methods. The variations listed in the academic liter-
methods [3,22,28,44,49,50]. The SEM selection problems depicted ature are generally concerned with the selection process in SEM
in Fig. 8 and Table 1 can be further visualised as Fig. 9. [17,19,27,81]. Examples of these data variations are shown in
The selection process shown in Fig. 8 can help to achieve the Fig. 12. Moreover, the data variations can be viewed as a special
aim of selecting the best method. Table 1 has shown the overall scenario to offer a clear discussion and analysis of this concern.
effectiveness of the selection challenge. In this context, on the basis For example, a scenario may occur as a goal maximisation rep-
of the SEM process involved in deriving the reliability and valid- resentative, which is behaviourally related to the SEM matrix. In
ity results, the visualisation in Fig. 9 can be used to further elabo- this context, the scenario can be observed when the maximisation
rate the aforementioned effectiveness. In particular, in the selection values of the criteria (↑) affect the data of the alternatives (high,
process associated with effectiveness (A) and (B) variables, each higher and highest levels), in which the aim is to achieve the bet-
output process can be considered an input to another process and ter selection process.
thus also involves an effect. From the discussions shared previously, the selection process
The process of SEM can be affected by two kinds of effective- therefore involves a simultaneous consideration of the multi-
ness variables, namely, (A) and (B). For effectiveness (A), the sup- criteria matrices of the SEM process, in which different maximi-
ported hypotheses are dependent on the multi-assessment of the sation goal scenarios are represented by the varying high, higher
measurement model affecting the variable selection. For effective- and highest levels that generate a variety of data.

8
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

Fig. 9. SEM effectiveness to ANN and the contribution process.

Fig. 10. Multi-criteria evaluation issues of the SEM processes.

Fig. 11. Criteria importance issues of the SEM processes.

7. Research proposals on potential future direction 12 illustrate the aforementioned multi-criteria evaluation issues.
For SEM, the experiments (Ei) are considered the available alter-
According to the analytic review of the literature regarding the natives; this approach is a suitable method of assessing reliability
performance of the SEM categories (Fig. 2) and the information and validity (multi-assessment of the measurement model) with
provided in Table 1, no exclusive study had used multi-criteria de- respect to the related criteria (R2 s, Q2 and F2 ). The adaptation of
cision making (MCDM) for different multi-assessment approaches precise and suitable techniques for selecting decisions by using the
to purposively ensure an easy evaluation of their respective ob- multi-criteria scheme can eventually boost decision-making qual-
jectives. Similarly, a comprehensive process is generally lacking for ity. Aside from the analysis, assessment and ranking shared previ-
selecting the most suitable method that can be used to assess the ously, MCDM is a good solution to aid decision makers in organis-
measurement model of the SEM method, which subsequently can ing and solving any problem [1].
affect the overall SEM processes. The present study is an attempt
to describe the potential future direction of the selection process
8. MCDM: an overview
(i.e. SEM method). This one-part selection process can be consid-
ered the main challenge in the literature, as the selection process
MCDM is an extension of decision theory that deals with de-
may be hindered by a multi-complex attribute problem. Figs. 10–
cisions that include multiple goals. The MCDM approach enables

9
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

Table 2
DM of the multi-assessment model in SEM.

SEM Criteria
Ei Alternatives R2 s Q2 F2

E1 R2 s/E1 Q2 / E1 F2 / E1
E2 R2 s/E2 Q2 / E1 F2 / E1
E3 R2 s/E3 Q2 / E1 F2 / E1
. . . .
. . . .
En R2 s/En Q2 / En F2 / En

R2 s = Coefficient of determination for SEM, Q2 = Predictive rele-


vance, F2 = Effect size, Ei= Experiment

it does not take into account their relative significance [158–162].


Thus, VIKOR is appropriate in terms of ranking conditions with
several alternatives and attributes [154,163,164]. VIKOR is one of
the most practical ways for solving real-world problems. The ma-
jor drawbacks of VIKOR include the absence of weight generation
provision [104,154,165]; by contrast, the analytic hierarchy process
(AHP) can provide a solution to such problems. The most recent
consideration for the MCDM methods is the combination of two or
more techniques to overcome the limitations of a single method-
ology. AHP and VIKOR are both frequently utilised MCDM tech-
niques in different research fields [154,166]. According to our anal-
ysis, AHP–VIKOR integration is the most appropriate MCDM solu-
tion. AHP can be utilised for weighting the multi-assessment crite-
ria. This phase intends to resolve the issue of significance criteria
by investigating the most affected criterion used in the SEM selec-
tion process. Then, VIKOR can be used to select the most suitable
method of assessing the measurement model of the SEM method
based on the weighted criteria resulting from the first stage. The
goal of this stage is to eliminate the issues of multi-criteria and
variations in the data. The complete methodology phases are illus-
trated in the next section.
Fig. 12. Variations of issues in the SEM process.

9. Methodology
alternatives to be assessed and analysed using several sets of
This section illustrates the proposed methodology of selecting
criteria, which are then combined to provide an overall assess-
the SEM methods. The methodology can be performed in three
ment [104–108]. Decision-making methods are well recognised,
phases. The flowchart depicting the methodology is illustrated in
and the MCDM is the most important of them. The method in-
Fig. 13.
cludes many processes, including organising, planning and resolv-
The first phase represents the identification and construction
ing distinct issue decisions by using a set of criteria [109,110]. This
phase of the SEM DM. In the second phase (development), the
technique incorporates different processes to improve the qual-
AHP–VIKOR integration is presented. In the third phase, the valida-
ity of decisions [111,112]. MCDM is accomplished by enabling a
tion approach is illustrated for validating the proposed work. The
more logical, efficient, clear and explicit decision-making process
details are provided in the following subsections.
compared with other conventional processes [113–117]. Therefore,
selecting the most appropriate MCDM technique is highly im-
portant. The mathematical model of MCDM can be represented 9.1. Identification and construction phase
as max[f1 (x ), f2 (x ), . . . , fn (x )], where n is the number of crite-
ria, and x ∈ A = [a1 , a2 , . . . , am ] [118–130]. The variable A repre- In this phase, the DMs of the SEM methods are identified and
sents the set of alternatives, and m is the number of alterna- conducted. The DM is considered to be the main component in the
tives. For each criterion j and alternative i, fij = fj (ai ), ∀(i, j ); i = proposed selection methodology, which is composed of decision al-
1, 2, . . . , m; j = 1, 2, . . . , n, which represent the dependencies of ternatives (i.e. set of experiment methods denoted by Ei) and three
each attribute value on the ith alternative and the jth criterion evaluation criteria (Table 2).
[131–140]. Various MCDM techniques have been developed and The most popular measure for evaluating the structural model
used in different areas [141–152] for weighting the evaluation cri- is R2 . Besides evaluating the values of R², F2 can be used to eval-
teria and alternative ranking. Ranking methods (e.g. technique for uate the amount of change in R2 when a specified exogenous
order of preference by similarity to the ideal solution [TOPSIS] construct is omitted from the model. Similarly, Q² is used by re-
and vlsekriterijumska optimizcija i kaompromisno resenje [VIKOR]) searchers to evaluate a model’s out-of-sample predictive power.
are used to determine the best alternative [153–155]. TOPSIS and The values can be obtained using statistical software, such as
VIKOR are suitable to situations involving several options and cri- SmartPLS. All of the evaluation criteria are regarded as benefit cri-
teria; these techniques are also particularly advantageous when teria, in which high values indicate better performance. However,
quantitative or objective data are provided [156,157]. The TOPSIS as discussed previously, the selection process of the SEM methods
approach finds the answer with the shortest distance to the ideal is a complex MCDM problem. To this end, the development of an
solution and the largest distance to the negative-ideal solution, but appropriate decision-making approach is an important preclusion

10
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

Fig. 13. Selection methodology of the multi-assessment model in SEM.

Fig. 14. Integration of the AHP–VIKOR methods.

of the problem complexity. The details are presented in the next 9.2.1. AHP method
subsection. In the AHP method [175], several steps are conducted for
weighting the SEM evaluation criteria.

9.2. Development Phase


Step 1: A pairwise comparison amongst the SEM evaluation
criteria is performed to extract the weights in a subjective
The development phase is based on the AHP method, which is
manner (Fig. 15).
implemented for the subjective weighting of the criteria used in
Step 2: The AHP constructs a pairwise matrix comparison based
the selection process of the SEM methods (Section 9.2.1). Then,
on Eq. (1) to provide a weighting decision [167].
VIKOR is used to rank and select the best SEM methods based
on the weighted criteria (Section 9.2.2). The MCDM contexts used ⎛ ⎞
X11 ··· X1n
to reach a final decision for the SEM selection are also presented
in this phase (Section 9.2.3). The structure of the integrated AHP– A = ⎝ ... ..
.
.. ⎠
. (1a)
VIKOR method is presented in Fig. 14. Xn1 ··· Xnn

11
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

Fig. 15. Pairwise comparison example.

1
W here Xii = 1, Xii = Xi j
9.2.2. VIKOR method
The VIKOR method is utilised to rank and select the most suit-
Step 3: A pairwise comparison questionnaire is designed for the able SEM method.
DM criteria of SEM. Then, the questionnaire is distributed to
experts. The target experts must have sufficient experience Step 1: Calculate the best f ∗ i and worst f − i values of the SEM
in the given field. Their judgments on the SEM evaluation evaluation criteria.
criteria must also be assessed. fi∗ = max j fi j, fi− = min j fi j (8)
Step 4: In this stage, the elements in matrix A (1) are nor-
malised to create the normalised matrix Anorm , Anorm (ai j) as where i = 1; 2; ...; n.
follows:
Xi j Step 2: As discussed previously, the AHP method is used
a i j = n (2) for weighting the SEM evaluation criteria. A collection
i=1 X i j of weights w = w1 , w2 , w3 , · · · , wj , · · · , wn is adopted in
⎛ ⎞
a11 ··· ain this step, and the result must be equal to 1. The resultant
= ⎝ ... .. ⎠
.. matrix is calculated using the following equation:
Anorm . . (3)
fi∗ − f i j
an1 ··· ann W M = wi∗ (9)
fi∗ − fi−
Step 5: In this step, the given pairwise comparisons are used
on the basis of mathematical calculations, and weights are Then, a weighted matrix is computed as follows:
assigned for the evaluation criteria used in the DM of SEM. ⎡ w1 ( f ∗ − f11 ) wi ( fi∗ − fi j )⎤
The weights of each criterion can be obtained according to
i
f ∗− f −
··· fi∗ − fi−
Eq. (4) [168]: ⎢. i i ⎥
n
⎢. .. .. ⎥ (10)

n ⎣. . . ⎦
j=1 ai j w1 ( fi∗ − f31 ) ( fi∗ − fi j )
Wi = and Wi = 1 (4) ···
wi
n fi∗ − fi− fi∗ − fi−
j=1

where n is the number of compared criteria. Step 3: In this step, the Sj and Rj values denoted by j=1,2,3,….,J,
i=1,2,3,…,n are calculated on the basis of the following equa-
Step 6: In this step, Eq. (5) is used to assess the CR with respect tions:
to the pairwise comparison matrix as follows:

n
fi∗ − fi j
CI Sj = wi∗ (11)
CR = (5) fi∗ − fi−
RI i=1

The consistency index (CI) is calculated according to Eq. (6) as fi∗ − f i j


follows: R j = maxi wi∗ (12)
fi∗ − fi−
λ max − n
CI = (6)
n−1 where wi is the weight of each criterion representing its relative
The random CI (RI) is calculated according to Eq. (7) as follows: significance.

1.98 (n − 1 ) Step 4: Calculate the values of Qj, j = (1, 2, · · · , J ) according to


RI = CI (7) the following equation:
n
   
A pairwise comparison matrix with a corresponding CR of less V S j − S∗ ( 1 − V ) R j − R∗
than 10% is agreeable; otherwise, the process must be repeated. Qj = + (13)
S− − S∗ R− − R∗
12
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

where 9.3. Validation phase


S∗ = min j S j , S− = max j S j R∗ = min j R j , R− = max j R j
The validity of any proposed work (e.g. method, technique,
Step 5: The SEM methods can finally be ranked. This process is model or framework) is a crucial step; this vital process is indis-
achieved by arranging the Q values in ascending order. The pensable when it comes to human life. The MCDM methods can
SEM method with the lowest Q value is considered to be the help to address SEM cases with respect to their multi-criteria com-
most suitable. plex problems that are further hindered by decision making. More
importantly, the development of a new decision support approach
9.2.3. MCDM contexts
cannot be applied unless validation is achieved. The validation in
Two different MCDM contexts are used in this study, with the
MCDM coincides with the principle of testing the reliability and
aim of reaching a final decision for the SEM selection.
stability of the decision results [170]. The literature shows sev-
A. The SEM selection process may include an individual con-
eral approaches for validating the proposed decision support ap-
text. Hence, the Kendall tau distance rank test is used to determine
proaches. One of these approaches is objective validation, which
whether the distance between each expert’s unique judgement is
uses mean ± standard deviation in statistical analysis, to validate
identical/acceptable. The Kendall tau distance rank introduced by
the ranking results of the proposed MCDM framework. In this ap-
Maurice Kendall is a measure for quantifying the pairwise discrep-
proach, the ranking results of the alternatives are divided into (n)
ancies between two ranking lists [169]. This non-parametric test
groups. Each group consists of a set of studied alternatives. The
can also quantify the relationships of two ranking lists. The corre-
mean ± standard deviation is computed for each group according
lation coefficient is preceded by a numeric value between 0 and 1,
to Eqs. 16 and 17.
where
1 n
• 0 indicates the existence of a complete connection between two mean = xi , (16)
n i=1
ranking lists (i.e. individual prioritisation of each expert), and

• 1 indicates the existence of a complete connection between two
1 N
ranking lists (i.e. individual prioritisation of each expert). Standard deviation = (xi − x̄ )2 . (17)
N−1 i=1
The greater the distance between two ranking lists, the stronger
Suppose we divided the rank results of the alternatives into
the connection between them. The following steps may be used to
three groups. According to this division, the best value should be
determine the Kendall tau rank distance:
for the first group, which must be supported by the mean ± stan-
Step 1: Create a table containing two ranking lists. Each expert’s dard deviation result. The statistical measurement of all of the
rating may be sorted in an ascending manner. remaining groups should be poorer than the first and preceding
Step 2: Calculate the number of concordant (C) pairings. In this groups; i.e. the obtained measurement should be better or equal
phase, the concordant pairings may be computed using the to those of the posterior groups, to ensure that the alternative pri-
rated alternatives of each expert. This process is used to de- oritisation has undergone systematic ranking [171].
termine the number of higher ranks that are below a given
level. 10. Potential future work
Step 3: Calculate the number of discordant (D) pairings. The
number of discordant pairings is expected to be identical to This section contains a discussion of potential future work for
that in the preceding stage, except that it entails the lower the proposed new context of the multi-assessment model in SEM
ranks instead of the higher ranks. based on MCDM combined with ANN. The findings from the com-
Step 4: Calculate the sums of the concordant and discordant prehensive research indicates that most of the studies had focused
pairings. on behavioural intention to use technology or apps in different sec-
Step 5: Calculate the degree of connection between the two tors. However, the research on interactive pretend play apps is lim-
ranking lists according to Eq. (15). ited. Hence, the first set of potential future work is in predicting
the factors that can influence the design of a low-cost autism chil-
C −− D
Kendal l  sT au = (15) dren’s app. In this context, the app can provide a better learning
C + D
environment and reduce the communication barriers to support
B. The term ‘Group MCDM’ refers to a scenario in which more children with autism to adapt to life and improve their living con-
than one DM is required to select the optimal alternative. G-MCDM ditions [172]. The second set of potential future work regarding the
techniques gather and integrate the knowledge and judgement of multi-assessment model in SEM based on MCDM combined with
experts from numerous areas in a systematic manner. In a group ANN is the prediction of factors that can influence the design of
setting, each expert contributes his or her subjective opinion to virtual reality games and augmented reality in pretend play ther-
the criteria that need subjective assessment. Numerous real-world apies. The reviewed literature indicates a lack of investigation of
decision-making issues encountered by organisations need collab- virtual reality and augmented reality with respect to the multi-
orative efforts amongst DM teams. The concept of group decision assessment model in SEM based on MCDM combined with ANN.
making refers to the combined results of many expert judgments By developing a training platform, the implementation of virtual
into a single and unique conclusion. Academic researchers of col- reality and augmented reality via the proposed method can help
lective decision-making have proposed two distinct configurations, children with autism to improve their communication and social
namely, internal aggregation and external aggregation [121]. Inter- skills with the people around them, thus enhancing their interac-
nal aggregation is aimed at aggregating the criteria weights from tion under different situations [173,174].
different experts into single unified weights by using the arith-
metic mean operator. Then, the unified weights of each evaluation 11. Conclusion
criteria are introduced into the VIKOR method to extract the inter-
nal group’s MCDM results. Meanwhile, the internal group’s MCDM This study delved into the communication workflow of the
results can be obtained by aggregating the score of each ranked multi-assessment model in SEM based on MCDM combined with
SEM method that has resulted from each individual ranking using ANN by implementing a systematic literature review. The underly-
the arithmetic mean operator. ing principles depicting the completeness of the SEM approaches

13
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

Acknowledgement

The authors are grateful to the Universiti Pendidikan Sultan


Idris, Malaysia for funding this study under UPSI SIG Grant No.
2020-0159-106-01.

References

[1] Chen H, Liu H, Chu X, Zhang L, Yan B. A two-phased SEM-neural network ap-
proach for consumer preference analysis. Advanced Engineering Informatics
2020;46:101156.
[2] Hew J-J, Badaruddin MN, Moorthy MK. Informatics, "Crafting a smartphone
repurchase decision making process: Do brand attachment and gender mat-
ter? Telematics and Informatics 2017;34(4):34–56.
[3] Hew J-J, Leong L-Y, Tan GW-H, Lee V-H, Ooi K-B. Mobile social tourism shop-
ping: A dual-stage analysis of a multi-mediation model. Tourism. Manage-
ment 2018;66:121–39.
[4] Li Y, Yang S, Zhang S, Zhang WJT. Informatics Mobile socialm edia use in-
tention in emergencies among Gen Y in China: An integrative framework of
gratifications, task-technology fit, and media dependency. Telematics and In-
formatics 2019;42:101244.
[5] Liébana-Cabanillas F, Marinkovic V, de Luna IR, Kalinic ZJ, Change S.
Predicting the determinants of mobile payment acceptance: A hybrid
SEM-neural network approach. Technological Forecasting and Social Change
2018;129:117–30.
[6] Leong L-Y, Hew T-S, Ooi K-B, Chong AY-L. Predicting the antecedents of trust
in social commerce–A hybrid structural equation modeling with neural net-
work approach. Journal of Business Research 2020;110:24–40.
[7] Leong L-Y, Hew T-S, Ooi K-B, Dwivedi YK. Predicting trust in online adver-
tising with an SEM-artificial neural network approach. Expert Systems with
Applications 2020;162:113849.
[8] Leong L-Y, Hew T-S, Ooi K-B, Wei J. Predicting mobile wallet resistance: A
two-staged structural equation modeling-artificial neural network approach.
International Journal of Information Management 2020;51:102047.
[9] Liébana-Cabanillas F, Marinković V, Kalinić Z. A SEM-neural network ap-
proach for predicting antecedents of m-commerce acceptance international.
Fig. 16. New context of the multi-assessment model in SEM based on MCDM com-
Journal of Information Management 2017;37(2):14–24.
bined with ANN.
[10] Song M, Qiao L, Law R. Engineering, "Formation path of customer engage-
ment in virtual brand community based on back propagation neural network
algorithm. International Journal of Computational Science and Engineering
2020;22(4):454–65.
can affect the selection guidelines either in the SEM method or in [11] Higueras-Castillo E, Kalinic Z, Marinkovic V, Liébana-Cabanillas FJ. A mixed
ANN and contribution approaches. Similarly, the existence of open analysis of perceptions of electric and hybrid vehicles. Energy Policy
issues and challenges related to the validity and reliability indi- 2020;136:111076.
[12] Ooi K-B, Lee V-H, Tan GW-H, Hew T-S, Hew J-J. Cloud computing in man-
cators in SEM proved the need to investigate the selection pro- ufacturing: the next industrial revolution in Malaysia? Expert Systems with.
cess. In this research, the selected cases highlighted the need for Applications 2018;93:376–94.
the SEM methods to be combined with the MCDM techniques, and [13] Chen C-C, Tsang S-S. Predicting adoption of mobile payments from the per-
spective of taxi drivers. IET Intelligent Transport Systems 2019;13(7):1116–24.
a new research methodology must emerge to support the selec-
[14] Abubakar AM, Namin BH, Harazneh I, Arasli H, Tunç T. Does gender mod-
tion process. Consequently, this study developed a novel integrated erates the relationship between favoritism/nepotism, supervisor incivility,
MCDM methodology that can be successfully linked with the SEM cynicism and workplace withdrawal: a neural network and SEM approach.
Tourism Management Perspectives 2017;23:129–39.
approaches in all of the three distinctly connected phases. In the
[15] Kim Y-J, Lee J-H, Lee S-W, Kim WY. Use of quick sequential organ failure as-
proposed solving methodology, MCDM plays a key role in the se- sessment score-based sepsis clinical decision support system may be helpful
lection process by overcoming the identified issues and reshaping to predict sepsis development. Journal of Anesthesia, Intensive Care, Emer-
the multi-assessment model in SEM combined with ANN, as previ- gency and Pain Medicine 2021;17(5):86–94 September 2021.
[16] Foo P-Y, Lee V-H, Tan GW-H, Ooi K-B. A gateway to realising sustainability
ously presented in Fig. 7. The new context of the multi-assessment performance via green supply chain management practices: a PLS–ANN ap-
model in SEM based on MCDM combined with ANN is shown in proach. Expert Systems with Applications 2018;107:1–14.
Fig. 16. This new representation can properly reproduce the selec- [17] Nair DJ, Rashidi TH, Dixit VV. Estimating surplus food supply for food rescue
and delivery operations. Socio-Economic Planning Sciences 2017;57:73–83.
tion of each identified alternative with respect to its related cri- [18] Priyadarshinee P, Raut RD, Jha MK, Gardas BB. Understanding and pre-
teria, thus overcoming the effectiveness problems of the SEM ap- dicting the determinants of cloud computing adoption: A two staged
proaches. Moreover, for the developed DM, the corresponding ef- hybrid SEM-Neural networks approach. Computers in Human Behavior
2017;76:341–62.
fectiveness problems can be solved, and many SEM reliability and [19] Raut RD, Mangla SK, Narwane VS, Gardas BB, Priyadarshinee P, Narkhede BE.
validity methods can be prioritised and selected. Consequently, the Linking big data analytics and operational sustainability practices for sustain-
appropriate method for obtaining superior and accurate measure- able business management. Journal of Cleaner Production 2019;224:10–24.
[20] Raut R., Priyadarshinee P., Gardas B. B., Narkhede B. E., Nehete R. J. B. A. I. J.
ment results can be accomplished. As a future direction, the test-
The incident effects of supply chain and cloud computing integration on the
ing results of the developed DM may be analysed under different business performance: An integrated SEM-ANN approach. Benchmarking: An
scenarios, followed by the validation of the results . International. Journal 2018;25(8):2688–2722.
[21] Najmi A., Kanapathy K., Aziz A. A. J. C. S. R., Management E. Exploring con-
sumer participation in environment management: Findings from two-staged
structural equation modelling-artificial neural network approach. Corporate
Social Responsibility and Environmental Management 2021;28(1):184–195.
Declaration of Competing Interest [22] Zabukovšek S. S., Kalinic Z., Bobek S., Tominc P. J. C. E. J. o. O. R. SEM–ANN
based research of factors’ impact on extended use of ERP systems. Central
The authors declare that they have no known competing finan- European Journal of Operations Research,vol 2019;27(3):703–735.
[23] Li X, et al. Eutrophication research of Dongting Lake: an integrated ML-SEM
cial interests or personal relationships that could have appeared to with neural network approach. International Journal of Environment and Pol-
influence the work reported in this paper. lution 2017;62(1):31–52.

14
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

[24] Sila I., Walczak S. J. P. P., Control Universal versus contextual effects on TQM: [50] Najmi A, Kanapathy K, Aziz AA, Management E. Exploring consumer par-
a triangulation study using neural networks. Production Planning & Control ticipation in environment management: findings from two-staged structural
2017;, vol. 28, no. 5,:367–386. equation modelling-artificial neural network approach. Corp Soc Responsib
[25] Sharma S. K., Al-Badi A., Rana N. P., Al-Azizi L. J. G. I. Q. Mobile applications Environ Manag 2021;28:184–95.
in government services (mG-App) from user’s perspectives: A predictive mod- [51] Abbas S, Hadi A-A, Abdullah HO, Alnoor A, Khattak ZZ, Khaw KW. Encoun-
elling approach. Government Information Quarterly 2018;35(4):557–568. tering Covid-19 and perceived stress and the role of a health climate among
[26] Ooi K.-B., Tan G. W.-H. J. E. S. w. A. Mobile technology acceptance model: An medical workers. Curr Psychol 2021:1–14.
investigation using mobile users to explore smartphone credit card. Expert [52] Al-Abrrow H, Ali J, Alnoor AJ. Multilevel influence of routine redesign-
Systems with Applications 2016;59:33–46. ing, legitimacy and functional affordance on sustainability accounting:
[27] Raut R. D., Priyadarshinee P., Gardas B. B., Jha M. K. J. T. F., Change S. An- mediating role of organizational sense-making. Global Business Review
alyzing the factors influencing cloud computing adoption using three stage 2019:0972150919869726.
hybrid SEM-ANN-ISM (SEANIS) approach. Technological Forecasting and So- [53] AL-Abrrow H, et al. Understanding employees’ responses to the COVID-19
cial Change 2018;134:98–123. pandemic: The attractiveness of healthcare jobs. Journal of Public Affairs
[28] Shahzad F, Xiu G, Khan MAS, Shahbaz MJ. Predicting the adoption of a mo- 2020;1-15.
bile government security response system from the user’s perspective: An [54] Hsu S-H, Chen W-h, Hsieh M-j, Excellence B. Robustness testing of PLS, LIS-
application of the artificial neural network approach. Technology in Society REL, EQS and ANN-based SEM for measuring customer satisfaction. Total
2020;62:101278. Quality Management & Business Excellence 2006;17(3):355–72.
[29] Islam AN, Laato S, Talukder S, Sutinen EJ, Change S. Misinformation sharing [55] Abdullah H, Ismail I, Alnoor A, Yaqoub EJ. Benchmarking Effect of perceived
and social media fatigue during COVID-19: An affordance and cognitive load support on employee’s voice behaviour through the work engagement: a
perspective. Technological Forecasting and Social Change 2020;159:120201. moderator role of locus of control. International Journal of Process Manage-
[30] Alkawsi GA, et al. A hybrid SEM-neural network method for identifying accep- ment and Benchmarking 2021;11(1):60–79.
tance factors of the smart meters in Malaysia: Challenges perspective. Alexan- [56] Cooper C, Booth A, Varley-Campbell J, Britten N, rGarside m. Defining the
dria Engineering Journal 2021;60(1):227–40. process to literature searching in systematic reviews: a literature review
[31] K. Batra, A. E. Morgan, M. J. J. o. A. Sharma, Intensive Care, and E. Medicine, of guidance and supporting studies. BMC Medical Research Methodology
"COVID-19 and social isolation endangering psychological health of older 2018;18(1):1–14.
adults: Implications for Telepsychiatry," Environmental and Occupational [57] Zaidan AA, et al. A survey on communication components for IoT-based tech-
Health, p. 1, 2020. nologies in smart homes. Telecommun Syst 2018;69(1):1–25.
[32] Sharma SK, Joshi A, Sharma HJ. A multi-analytical approach to predict [58] Mohsin A, et al. Real-time remote health monitoring systems using body sen-
the Facebook usage in higher education. Computers in Human Behavior sor information and finger vein biometric verification: A multi-layer system-
2016;55:340–53. atic review. J Med Syst 2018;42(12):1–36.
[33] Tiruwa A, Yadav R, Suri PK. Modelling Facebook usage for collaborative learn- [59] Mohsin A, et al. Real-time medical systems based on human bio-
ing in higher education. Journal of Applied Research in Higher Education metric steganography: A systematic review. J Med Syst 2018;42(12):1–
2018;10(3):12–45. 20.
[34] Binsawad M. Corporate Social Responsibility in Higher Education: A PLS-SEM [60] Talal M, et al. Smart home-based IoT for real-time and secure remote health
Neural Network Approach. IEEE Access 2020;8:29125–31. monitoring of triage and priority system using body sensors: multi-driven
[35] Kalinic Z, Marinkovic V, Molinillo S, Liebana-Cabanillas FJ, Services C. A mul- systematic review. J Med Syst 2019;43(3):42.
ti-analytical approach to peer-to-peer mobile payment acceptance predic- [61] Mohsin A, et al. Blockchain authentication of network applications: Tax-
tion. Journal of Retailing and Consumer Services, Elsevier 2019;49(C):143– onomy, classification, capabilities, open challenges, motivations, recommen-
153. dations and future directions. Comput Standards Interfaces 2019;64:41–
[36] Sharma SK, Sharma MJ. Examining the role of trust and quality dimensions 60.
in the actual usage of mobile banking services: An empirical investigation. [62] Mohsin A, et al. Based medical systems for patient’s authentication: to-
International Journal of Information Management 2019;44:65–75. wards a new verification secure framework using CIA standard. J Med Syst
[37] Coelho A, Moutinho L, Hutcheson GD, Santos Silva MM. Artificial neural net- 2019;43(7):1–34.
works and structural equation modelling: an empirical comparison to evalu- [63] Mohsin A, et al. Finger vein biometrics: taxonomy analysis, open challenges,
ate business customer loyalty. Quantitative modelling in marketing and man- future directions, and recommended solution for decentralised network ar-
agement. World Scientific; 2013. p. 117–49. chitectures. IEEE Access 2020;8:9821–45.
[38] Pičuljan A, Protić A, Haznadar M, Šustić AJSV. The role of B-line artifacts on [64] Shuwandy ML, et al. mHealth authentication approach based 3D touchscreen
lung ultrasound in critically ill patients. Journal of Anesthesia, Intensive Care, and microphone sensors for real-time remote healthcare monitoring system:
Emergency and Pain Medicine 2020;7:1. comprehensive review, open issues and methodological aspects. Comput Sci
[39] Alam MZ, Hu W, Kaium MA, Hoque MR, Alam M. Understanding the deter- Rev 2020;38:10 030 0.
minants of mHealth apps adoption in Bangladesh: a SEM-Neural network ap- [65] Shuwandy ML, Zaidan B, Zaidan A, Albahri AS. Sensor-based mHealth authen-
proach. Technology in Society 2020;61:101255. tication for real-time remote healthcare monitoring system: a multilayer sys-
[40] M. S. Talukder, G. Sorwar, Y. Bao, J. U. Ahmed, M. A. S. J. T. F. Palash, and tematic review. J Med Syst 2019;43(2):33.
S. Change, "Predicting antecedents of wearable healthcare technology accep- [66] Alamoodi A, et al. Sentiment analysis and its applications in fighting
tance by elderly: a combined SEM-Neural Network approach," Technological COVID-19 and infectious diseases: a systematic review. Expert Syst Appl
Forecasting and Social Change, vol. 150, p. 119793, 2020. 2020:114155.
[41] Asadi S., Abdullah R., Safaei M., Nazir S. J. M. I. S. An integrated SEM- [67] Albahri AS, et al. IoT-based telemedicine for disease prevention and health
neural network approach for predicting determinants of adoption of wear- promotion: State-of-the-Art. J Netw Comput Appl 2021;173:102873.
able healthcare devices," Wearable Technology and Mobile Applications for. [68] Hamid RA, et al. How smart is e-tourism? A systematic review of smart
Healthcare 2019;2649:34–50. tourism recommendation system applying data management. Comput Sci Rev
[42] Dadashova B, Arenas-Ramírez B, Mira-McWilliams J, Aparicio-Izquierdo F. 2021;39:100337.
Prevention, “Methodological development for selection of significant predic- [69] Albahri AS, et al. Role of biological data mining and machine learning tech-
tors explaining fatal road accidents. Accid. Anal Prev 2016;90:82–94. niques in detecting and diagnosing the novel coronavirus (COVID-19): a sys-
[43] C. Parsad, S. Mittal, R. J. I. J. o. P. Krishnankutty, and P. Management, tematic review. J Med Syst 2020;44:1–11.
"A study on the factors affecting household solar adoption in Kerala, In- [70] Hair Jr JF, Sarstedt M, Hopkins L, Kuppelwieser VG. Partial least squares struc-
dia,"International Journal of Productivity and Performance Management, tural equation modeling (PLS-SEM): An emerging tool in business research.
2020, Vol 10, Issue 12, 23-33. European Business Review 2014;26(2):106–21.
[44] Wong L-W, Leong L-Y, Hew J-J, Tan GW-H, Ooi K-B. Time to seize the digital [71] Hair J, Hollingsworth CL, Randolph AB, Chong AY, Systems D. An updated and
evolution: adoption of blockchain in operations and supply chain manage- expanded assessment of PLS-SEM in information systems research. Industrial
ment among Malaysian SMEs. International Journal of Information Manage- Management & Data Systems 2017;117(3).
ment 2020;52:101997. [72] Henseler J, Ringle CM. A new criterion for assessing discriminant validity in
[45] Ahani A, Rahim NZA, Nilashi M. Forecasting social CRM adoption in SMEs: variance-based structural equation modeling. Journal of the Academy of Mar-
A combined SEM-neural network method. Computers in Human Behavior keting Science 2015;43(1):115–35.
2017;75:560–78. [73] Hair Jr JF, Howard MC, Nitzl C. Assessing measurement model quality in
[46] Lee V.-H., Foo A. T.-L., Leong L.-Y., Ooi K.-B. J. E. S. w. A. Can competitive PLS-SEM using confirmatory composite analysis. Journal of Business Research
advantage be achieved through knowledge management? A case study on 2020;109:101–10.
SMEs,"Expert Systems with Applications 2016;65:136–151. [74] Gerbing DW, Hamilton JG. Viability of exploratory factor analysis as a precur-
[47] Khayer A, Talukder MS, Bao Y, Hossain MN. Cloud computing adoption and its sor to confirmatory factor analysis. Structural Equation Modeling. A Multidis-
impact on SMEs’ performance for cloud supported operations: a dual-stage ciplinary Journal 1996;3(1):62–72.
analytical approach. Technology in Society 2020;60:101225. [75] Kalinić Z, Marinković V, Kalinić L, Liébana-Cabanillas FJ. Neural network mod-
[48] Alnoor A. Human capital dimensions and firm performance, mediating role eling of consumer satisfaction in mobile commerce: an empirical analysis. Ex-
of knowledge management. International Journal of Business Excellence pert Systems with Applications 2021;175:114803.
2020;20(2):149–68. [76] Kheirollahpour MM, Danaee MM, Merican AFA, Shariff AA. Prediction of the
[49] Alam MZ, Hu W, Kaium A, Hoque R, Alam MMD. Understanding the deter- influential factors on eating behaviors: a hybrid model of structural equation
minants of mHealth apps adoption in Bangladesh: a SEM-Neural network ap- modelling-artificial neural networks. The Scientific World Journal 2020:1–12.
proach. Technology in Society 2020:101255. [77] Ray A, Bala PK, Rana NP. Exploring the drivers of customers’ brand attitudes

15
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

of online travel agency services: A text-mining based approach. Journal of [105] Yas QM, Zadain A, Zaidan B, Lakulu M, Rahmatullah B. Towards on develop a
Business Research 2021;128:391–404. framework for the evaluation and benchmarking of skin detectors based on
[78] Kalinic Z, Marinkovic V, Molinillo S, Liébana-Cabanillas FJ, Services C. A mul- artificial intelligent models using multi-criteria decision-making techniques.
ti-analytical approach to peer-to-peer mobile payment acceptance prediction. Int J Pattern Recognit Artif Intell 2017;31(03):1759002.
Journal of Retailing and Consumer Services 2019;49:143–53. [106] Abdullateef BN, Elias NF, Mohamed H, Zaidan A, Zaidan B. An evalua-
[79] Lee V-H, Hew J-J, Leong L-Y, Tan GW-H, Ooi K-B. Wearable payment: A deep tion and selection problems of OSS-LMS packages. SpringerPlus 2016;5(1):1–
learning-based dual-stage SEM-ANN analysis. Expert Systems with Applica- 35.
tions 2020;157:113477. [107] Zaidan B, Zaidan A, Karim HA, Ahmad NN. A new digital watermarking eval-
[80] Najmi A, Kanapathy K, Aziz A, Research P. Understanding consumer partic- uation and benchmarking methodology using an external group of evaluators
ipation in managing ICT waste: Findings from two-staged Structural Equa- and multi-criteria analysis based on ‘large-scale data. Software Practice and
tion Modeling–Artificial Neural Network approach. Environmental Science Experience 2017;47(10):1365–92.
and Pollution Research 2021;28(12):14782–96. [108] Zaidan B, Zaidan A. Software and hardware FPGA-based digital watermarking
[81] Chen H, Liu H, Chu X, Zhang L, Yan B. A two-phased SEM-neural network ap- and steganography approaches: toward new methodology for evaluation and
proach for consumer preference analysis. Advanced Engineering Informatics benchmarking using multi-criteria decision-making techniques. J Circuits, Syst
2020;46:101156. Comput 2017;26(07):1750116.
[82] Hew J-J, Badaruddin MN, Moorthy MK. Crafting a smartphone repurchase de- [109] Qader M, Zaidan B, Zaidan A, Ali S, Kamaluddin M, Radzi W. A methodology
cision making process: Do brand attachment and gender matter? Telematics for football players selection problem based on multi-measurements criteria
and. Informatics 2017;34(4):34–56. analysis. Measurement 2017;111:38–50.
[83] Alkawsi GA, et al. A hybrid SEM-neural network method for identifying ac- [110] Rahmatullah B, Zaidan A, Mohamed F, Sali A. Multi-complex attributes anal-
ceptance factors of the smart meters in Malaysia: Challenges perspective. AEJ ysis for optimum GPS baseband receiver tracking channels selection. In: 2017
- Alexandria Engineering Journal 2020:23–40. 4th international conference on control, decision and information technolo-
[84] Raut R, Priyadarshinee P, Gardas BB, Narkhede BE, Nehete RJ. The incident gies (CoDIT). IEEE; 2017. p. 1084–8.
effects of supply chain and cloud computing integration on the business per- [111] Zaidan B, Zaidan A, Abdul Karim H, Ahmad N. A new approach based on mul-
formance. Benchmarking An International Journal 2018;252:20–33. ti-dimensional evaluation and benchmarking for data hiding techniques. Int J
[85] Duan SX, Deng HJ, Systems D. Hybrid analysis for understanding contact trac- Inform Technol Decision Making 2017:1–42.
ing apps adoption. Industrial Management & Data Systems 2021:40–55. [112] Jumaah F, Zaidan A, Zaidan B, Bahbibi R, Qahtan M, Sali A. Technique for
[86] Asadi S, Nilashi M, Samad S, Rupani PF, Kamyab H, Abdullah RJ. A proposed order performance by similarity to ideal solution for solving complex situa-
adoption model for green IT in manufacturing industries. Journal of Cleaner tions in multi-criteria optimization of the tracking channels of GPS baseband
Production 2021;297:126629. telecommunication receivers. Telecommun Syst 2018;68(3):425–43.
[87] Talwar M, Talwar S, Kaur P, Tripathy N, Dhir AJ, Services C. Has financial at- [113] Almahdi E, Zaidan A, Zaidan B, Alsalem M, Albahri Albahri Ooms. Mo-
titude impacted the trading activity of retail investors during the COVID-19 bile-based patient monitoring systems: a prioritisation framework us-
pandemic? Journal of Retailing and Consumer Services 2021;58:102341. ing multi-criteria decision-making techniques. Journal of Medical Systems
[88] Kunnapapdeelert S, Pitchayadejanant KJ, Modelling SC. Hybrid SEM-neu- 2019;43(7):219.
ral networks for predicting electronics logistics information system adop- [114] Khatari M, Zaidan A, Zaidan B, Albahri OS, Alsalem M. Multi-criteria evalua-
tion in Thailand healthcare supply chain. Inderscience Enterprises Ltd tion and benchmarking for active queue management methods: Open issues,
2020;11(1):54–68. challenges and recommended pathway solutions. Int J Inform Technol Deci-
[89] Hayat N, Al Mamun A, Nasir NAM, Selvachandran G, Nawi NBC, Gai QS. Pre- sion Making 2019;18(04):1187–242.
dicting sustainable farm performance—using hybrid structural equation mod- [115] Mohammed R, et al. Determining importance of many-objective optimisation
elling with an artificial neural network approach. Land 2020;9(9):289. competitive algorithms evaluation criteria based on a novel fuzzy-weighted
[90] Shukla SJE. Technologies I. M-learning adoption of management students’: zero-inconsistency method. Int J Inform Technol Decision Making 2021:1–
a case of India. Education and Information Technologies 2021;26(1):279– 47.
310. [116] Mohsin A, et al. PSO–Blockchain-based image steganography: towards a new
[91] Liébana F, Cabanillas NSingh, Kalinic Z, Carvajal-Trujillo EJ. and Management, method to secure updating and sharing COVID-19 data in decentralised hos-
"Examining the determinants of continuance intention to use and the moder- pitals intelligence architecture. Multimedia Tools Appl 2021;80(9):14137–61.
ating effect of the gender and age of users of NFC mobile payments: a mul- [117] Krishnan E, et al. Interval type 2 trapezoidal-fuzzy weighted with zero incon-
ti-analytical approach. Information Technology and Management 2021:1–29 sistency combined with VIKOR for evaluating smart e-tourism applications.
volume. Int J Intell Syst 2021.
[92] Alam MMD, Alam MZ, Rahman SA, Taghizadeh SK. Factors influencing [118] Salman OH, Zaidan A, Zaidan B, NaserkalidHashim MJ, Making D. Novel
mHealth adoption and its impact on mental well-being during COVID-19 pan- methodology for triage and prioritizing using “big data” patients with
demic: A SEM-ANN approach. J Biomed Inform 2021;116:103722. chronic heart diseases through telemedicine environmental. International.
[93] Abbasi GA, Tiew LY, Tang J, Goh Y-N, Thurasamy RJ. The adoption of cryp- Journal of Information Technology & Decision Making 2017;16(05):1211–
tocurrency as a disruptive force: Deep learning-based dual stage struc- 1245.
tural equation modelling and artificial neural network analysis. PLOS one [119] Yas QM, Zaidan A, Zaidan B, Rahmatullah B, Karim HA. Comprehensive in-
2021;16(3):e0247582. sights into evaluation and benchmarking of real-time skin detectors: Re-
[94] Sharma A. A Structural equation modelling & artificial neural network ap- view, open issues & challenges, and recommended solutions. Measurement
proach to examine the impact of human resources accounting on firm’s per- 2018;114:243–60.
formance. Pacific Bus Rev Int Mar 2020;12(9):1–12. [120] Zaidan B, Zaidan A. Comparative study on the evaluation and benchmarking
[95] Kruk M, Paturej E, Artiemjew PJ. From explanatory to predictive network information hiding approaches based multi-measurement analysis using TOP-
modeling of relationships among ecological indicators in the shallow tem- SIS method with different normalisation, separation and context techniques.
perate lagoon. Ecological Indicators 2020;117:106637. Measurement 2018;117:277–94.
[96] ŞAHİN H, Topal BJ. Impact of information technology on business perfor- [121] Kalid N, et al. Based on real time remote health monitoring systems: a
mance: integrated structural equation modeling and artificial neural network new approach for prioritization “large scales data” patients with chronic
approach. Scientia Iranica 2018;25(3):1272–80. heart diseases using body sensors and communication technology. J Med Syst
[97] Sarstedt M, Ringle CM, Smith D, Reams R, Hair JF Jr. Partial least squares 2018;42(4):69.
structural equation modeling (PLS-SEM): A useful tool for family business re- [122] Zaidan A, et al. A review on smartphone skin cancer diagnosis apps in evalu-
searchers. Journal of Family. Business Strategy 2014;5(1):105–15. ation and benchmarking: coherent taxonomy, open issues and recommenda-
[98] Hair Jr JF, Sarstedt M, Hopkins L, Kuppelwieser VG. Partial least squares struc- tion pathway solution. Health and Technology 2018;8(4):223–38.
tural equation modeling (PLS-SEM). European Business Review 2014:40–60. [123] Albahri O, et al. Systematic review of real-time remote health monitoring
[99] Aria M, Cuccurullo CJ. bibliometrix: An R-tool for comprehensive science system in triage and priority-based sensor technology: Taxonomy, open chal-
mapping analysis. Journal of Informetrics 2017;11(4):959–75. lenges, motivation and recommendations. J Med Syst 2018;42(5):80.
[100] Almahdi E, Zaidan A, Zaidan B, Alsalem M, Albahri OS. Mobile patient moni- [124] Alsalem M, et al. Systematic review of an automated multiclass detection and
toring systems from a benchmarking aspect: Challenges, open issues and rec- classification system for acute Leukaemia in terms of evaluation and bench-
ommended solutions. Journal of Medical Systems 2019;43(7):1–23. marking, open challenges, issues and methodological aspects. J. Med Syst
[101] Zaidan AA, Zaidan BB, Al-Haiqi A, Kiah MLM, Hussain M, Abdulnabi MJ. Eval- 2018;42(11):204.
uation and selection of open-source EMR software packages based on inte- [125] Kalid N, Zaidan A, Zaidan B, Salman OH, Hashim M, Muzammil HJ. "Based real
grated AHP and TOPSIS. Journal of Biomedical Informatics 2015;53:390–404. time remote health monitoring systems: a review on patients prioritization
[102] Abdulkareem KH, et al. A novel multi-perspective benchmarking framework and related" big data". using body sensors information and communication
for selecting image dehazing intelligent algorithms based on BWM and group technology 2018;42(2):30 PMID: 29288419.
VIKOR techniques. International Journal of Information Technology & Decision [126] Jumaah F, Zadain A, Zaidan B, Hamzah A, Bahbibi RJM. Decision-making so-
Making 2020;19(03):909–57. lution based multi-measurement design parameter for optimization of GPS
[103] Albahri A, Hamid RA, Albahri O, Zaidan AJ. Detection-based prioritisation: receiver tracking channels in static and dynamic real-time positioning multi-
Framework of multi-laboratory characteristics for asymptomatic COVID-19 path environment. J Med Syst 2018;118:83–95.
carriers based on integrated Entropy–TOPSIS methods. Artificial Intelligence [127] Albahri A, Zaidan A, Albahri O, Zaidan B, Alsalem M. Real-time fault-tolerant
in Medicine 2020;111:101983. mHealth system: Comprehensive review of healthcare services, opens issues,
[104] Albahri O, et al. Fault-tolerant mHealth framework in the context of IoT-based challenges and methodological aspects. J Med Syst 2018;42(8):137.
real-time wearable health data sensors. IEEE Access 2019;7:50052–80. [128] Albahri O, Zaidan A, Zaidan B, Hashim M, Albahri A, Alsalem M. Real-time

16
A.S. Albahri, A. Alnoor, A.A. Zaidan et al. Chaos, Solitons and Fractals 153 (2021) 111445

remote health-monitoring Systems in a Medical Centre: A review of the pro- [154] Albahri A, et al. Based multiple heterogeneous wearable sensors: A smart re-
vision of healthcare services-based body sensor information, open challenges al-time health monitoring structured for hospitals distributor, 7. IEEE Access
and methodological aspects. J Med Syst 2018;42(9):164. 2019:37269–323.
[129] Zughoul O, et al. Comprehensive insights into the criteria of student perfor- [155] Albahri OS, et al. Multidimensional benchmarking of the active queue man-
mance in various educational domains. IEEE Access 2018;6(4):73245–64. agement methods of network congestion control based on extension of fuzzy
[130] Salih MM, Zaidan B, Zaidan A, Ahmed MA. Survey on fuzzy TOP- decision by opinion score method. Int J Intell Syst 2021;36(2):796–831.
SIS state-of-the-art between 2007 and 2017. Comput Operations Res [156] Dawood KA, Zaidan A, Sharif KY, Ghani AA, Zulzalil H, Zaidan B. Novel multi-
2019;104:207–27. -perspective usability evaluation framework for selection of open source soft-
[131] Albahri A, et al. Based multiple heterogeneous wearable sensors: A smart ware based on BWM and group VIKOR techniques. Int J Inform Technol Deci-
real-time health monitoring structured for hospitals distributor. IEEE Access sion Making 2021:1–91.
2019;7:37269–323. [157] Mohammed TJ, et al. Convalescent-plasma-transfusion intelligent framework
[132] Albahri O, et al. Fault-tolerant mHealth framework in the context of IoT-based for rescuing COVID-19 patients across centralised/decentralised telemedicine
real-time wearable health data sensors. IEEE Access 2019;7:50052–80. hospitals based on AHP-group TOPSIS and matching component. Appl Intell
[133] Almahdi E, Zaidan A, Zaidan B, Alsalem M, Albahri O, Albahri A. Mobile pa- 2021;51(5):2956–87.
tient monitoring systems from a benchmarking aspect: Challenges, open is- [158] Opricovic S, Tzeng O. Extended VIKOR method in comparison with outranking
sues and recommended solutions. J Med Syst 2019;43(7):207. methods. European Journal of Operational Research 2007;178(2):514–29.
[134] Alsalem M, et al. Multiclass benchmarking framework for automated acute [159] Salih MM, Zaidan B, Zaidan A. Fuzzy decision by opinion score method. Appl
Leukaemia detection and classification based on BWM and group-VIKOR. J Soft Comput 2020;96:106595.
Med Syst 2019;43(7):212. [160] Mohammed R, et al. Review of the research landscape of multi-criteria evalu-
[135] Almahdi E, Zaidan A, Zaidan B, Alsalem M, Albahri O, Albahri A. Mobile-based ation and benchmarking processes for many-objective optimization methods:
patient monitoring systems: A prioritisation framework using multi-criteria coherent taxonomy, challenges and recommended solution. Int J Inform Tech-
decision-making techniques. J Med Syst 2019;43(7):219. nol Decision Making (IJITDM) 2020;19(06):1619–93.
[136] Mohammed K, et al. Real-time remote-health monitoring systems: a review [161] Malik R, et al. Novel roadside unit positioning framework in the context of
on patients prioritisation for multiple-chronic diseases, taxonomy analysis, the vehicle-to-infrastructure communication system based on AHP—Entropy
concerns and solution procedure. J Med Syst 2019;43(7):223. for weighting and borda—VIKOR for uniform ranking. Int J Inform Technol
[137] Khatari M, Zaidan A, Zaidan B, Albahri O, Alsalem M. Multi-criteria evaluation Decision Making 2021:1–34.
and benchmarking for active queue management methods: open issues chal- [162] Khatari M, Zaidan A, Zaidan B, Albahri O, Alsalem M, Albahri A. Multidi-
lenges and recommended pathway solutions. Int J Inf Technol Decis Making mensional benchmarking framework for AQMs of network congestion con-
2019;18(4):1187–242. trol based on AHP and Group-TOPSIS. Int J Inform Technol Decision Making
[138] Alaa M, et al. Assessment and ranking framework for the English skills of 2021:1–38.
pre-service teachers based on fuzzy Delphi and TOPSIS methods. IEEE Access [163] Albahri AS, et al. Multi-biological laboratory examination framework for the
2019;7:126201–23. prioritization of patients with COVID-19 based on integrated AHP and group
[139] Ibrahim N, et al. Multi-criteria evaluation and benchmarking for young learn- VIKOR methods. Int J Inform Technol Decision Making 2020;19(05):1247–69.
ers’ English language mobile applications in terms of LSRW skills. IEEE Access [164] Salih MM, Albahri O, Zaidan A, Zaidan B, Jumaah F, Albahri A. Benchmarking
2019;7(7):146620–51. of AQM methods of network congestion control based on extension of inter-
[140] Talal M, et al. Comprehensive review and analysis of anti-malware apps for val type-2 trapezoidal fuzzy decision by opinion score method. Telecommun
smartphones. Telecommun Syst 2019;72(2):285–337. Syst 2021:1–30.
[141] Napi NM, Zaidan AA, Zaidan BB, Albahri OS, Alsalem MA, Albahri AS. Medical [165] Khatari M, Zaidan A, Zaidan B, Albahri O, Alsalem MJ. Multi-criteria evalua-
emergency triage and patient prioritisation in a telemedicine environment: a tion and benchmarking for active queue management methods: Open issues
systematic review. Health Technol 2019;9(5):679–700 2019/11/01. challenges and recommended pathway solutions. International Journal of In-
[142] Enaizan O, et al. Electronic medical record systems: Decision support exam- formation Technology and Decision Making 2019;18(4):1187–242.
ination framework for individual, security and privacy concerns using multi- [166] Albahri O, et al. New mHealth hospital selection framework supporting
-perspective analysis. Health Technol 2020;10(3):795–822. decentralised telemedicine architecture for outpatient cardiovascular dis-
[143] Zaidan A, Zaidan B, Alsalem M, Albahri O, Albahri A, Qahtan M. Multi-agent ease-based integrated techniques: Haversine-GPS and AHP-VIKOR. J Ambient
learning neural network and Bayesian model for real-time IoT skin detec- Intell Humanized Comput 2021:1–21.
tors: a new evaluation and benchmarking methodology. Neural Comput Appl [167] Baffoe GJE. planning p. Exploring the utility of Analytic Hierarchy Process
2020;32(12):8315–66. (AHP) in ranking livelihood activities for effective and sustainable rural devel-
[144] Tariq I, et al. MOGSABAT: a metaheuristic hybrid algorithm for solving multi- opment interventions in developing countries. Evaluation and Program Plan-
-objective optimisation problems. Neural Comput Appl 2018;32:2020. ning 2019;72:197–204.
[145] Zughoul O. Novel triplex procedure for ranking the ability of software engi- [168] Roy J, Chatterjee K, Bandyopadhyay A, Kar SJES. Evaluation and selection of
neering students based on two levels of AHP and group TOPSIS techniques. medical tourism sites: A rough analytic hierarchy process based multi-at-
Int J Inform Technol Decision Making 2020. tributive border approximation area comparison approach. Expert Systems
[146] Abdulkareem KH, et al. A new standardisation and selection framework for 2018;35(1):e12232.
real-time image dehazing algorithms from multi-foggy scenes based on fuzzy [169] Desarkar MS, Sarkar S, Mitra PJ. Preference relations based unsupervised rank
Delphi and hybrid multi-criteria decision analysis methods. Neural Comput aggregation for metasearch. Expert Systems with Applications 2016;49:86–98.
Appl 2020 2020/05/26. [170] Wu W, Xu Z. Hybrid TODIM method with crisp number and probabil-
[147] Mohammed K, et al. Novel technique for reorganisation of opinion order ity linguistic term set for urban epidemic situation evaluation. Complexity
to interval levels for solving several instances representing prioritisation in 2020;2020.
patients with multiple chronic diseases. Comput Methods Programs Biomed [171] Albahri O, et al. Systematic review of artificial intelligence techniques in
2020;185:105151. the detection and classification of COVID-19 medical images in terms of
[148] Mohammed K, et al. A Uniform Intelligent Prioritisation for Solving Diverse evaluation and benchmarking: taxonomy analysis, challenges, future solu-
and Big Data Generated From Multiple Chronic Diseases Patients Based on tions and methodological aspects. J Infect Public Health 2020;13(10):1381–96.
Hybrid Decision-Making and Voting Method. IEEE Access 2020;8:91521–30. doi:10.1016/j.jiph.2020.06.028.
[149] Zaidan A, Zaidan B, Alsalem M, Momani F, Zughoul O. Novel Multiperspective [172] Soomro N, Soomro SJ. Autism Children’s App using PECS. Annals of Emerging
Hiring Framework for the Selection of Software Programmer Applicants Based Technologies in Computing (AETiC). Vol. 2018;2(1):1–10.
on AHP and Group TOPSIS Techniques. Int J Inform Technol Decision Making [173] Syahputra M, et al. Implementation of augmented reality in pretend play
2020;18(4):1–73. therapy for children with autism spectrum disorder Journal of Physics: Con-
[150] Abdulkareem KH, et al. A novel multi-perspective benchmarking framework ference Series, 1235. IOP Publishing; 2019.
for selecting image dehazing intelligent algorithms based on bwm and group [174] Rahmadiva M, Arifin A, Fatoni MH, Baki SH, Watanabe T. A design of mul-
vikor techniques. Int J Inform Technol Decision Making 2020;19(3):909–57. tipurpose virtual reality game for children with autism spectrum disorder.
[151] Alamoodi A, et al. A systematic review into the assessment of medical apps: In: 2019 international biomedical instrumentation and technology conference
motivations, challenges, recommendations and methodological aspect. Health (IBITeC), 1. IEEE; 2019. p. 1–6.
Technol 2020:1–17. [175] Hamid RA, Albahri AS, Albahri OS, Zaidan AA. Dempster–Shafer theory for
[152] Albahri OS, et al. Helping doctors hasten COVID-19 treatment: Towards a res- classification and hybridised models of multi-criteria decision analysis for pri-
cue framework for the transfusion of best convalescent plasma to the most oritisation: a telemedicine framework for patients with heart diseases. J Am-
critical patients based on biological requirements via ml and novel MCDM bient Intell Humanized Comput 2021:1–35.
methods. Comput Methods Programs Biomed 2020;196:105617.
[153] Salih MM, Zaidan B, Zaidan A, Ahmed MA, Research O. Survey on fuzzy TOP-
SIS state-of-the-art between 2007 and 2017. Computers & Operations Re-
search 2019;104:207–27.

17

You might also like