Predictors of Smartphone Addiction and Social Isolation among Jordanian Children and Adolescents Using SEM and ML

: Smartphone addiction has become a major problem for everyone. According to recent studies, a considerable number of children and adolescents are more attracted to smartphones and exhibit addictive behavioral indicators, which are emerging as serious social problems. The main goal of this study is to identify the determinants that inﬂuence children’s smartphone addiction and social isolation among children and adolescents in Jordan. The theoretical foundation of this study model is based on constructs adopted from the Technology Acceptance Model (TAM) (i.e., perceived ease of use and perceived usefulness), with social inﬂuence and trust adopted from the TAM extended model along with perceived enjoyment. In terms of methodology, the study uses data from 511 parents who responded via convenient sampling, and the data was collected via a survey questionnaire and used to evaluate the research model. To test the study hypotheses, the empirical validity of the research model was set up, and the data were analyzed with SPSS version 21.0 and AMOS 26 software. Structural equation modeling (SEM), conﬁrmatory factor analysis (CFA), and machine learning (ML) methods were used to test the study hypotheses and validate the properties of the instrument items. The ML methods used are support vector machine (SMO), the bagging reduced error pruning tree (REPTree), artiﬁcial neural network (ANN), and random forest. Several major ﬁndings were indicated by the results: perceived usefulness, trust, and social inﬂuence were signiﬁcant antecedent behavioral intentions to use the smartphone. Also, ﬁndings prove that behavioral intention is statistically supported to have a signiﬁcant inﬂuence on smartphone addiction. Furthermore, the ﬁndings conﬁrm that smartphone addiction positively inﬂuences social isolation among Jordanian children and adolescents. Yet, perceived ease of use and perceived enjoyment did not have a signiﬁcant effect on behavioral intention to use the smartphone among Jordanian children and adolescents. The research contributes to the body of knowledge and literature by empirically examining and theorizing the implications of smartphone addiction on social isolation. Further details of the study contribution, as well as research future directions and limitations, are presented in the discussion section.


Introduction
In developed and developing countries, mobile phone use has changed over the past several decades, and smartphone addiction has emerged as a serious problem in societies. The Technology Acceptance Model (TAM) proposed by Davis et al. [34] is the most influential research model that is used to explain, determine, and understand the recent technology innovation adoption behavior of an individual. Usefulness, ease of use, and behavioral intention are used to assess and evaluate the actual adoption of recent technology, where ease of use and usefulness are the most important TAM determinants that predict individuals' behavior toward acceptance of new technology [34]. Davis et al. [34] stated that perceived ease of use and perceived usefulness have a direct influence on a The Technology Acceptance Model (TAM) proposed by Davis et al. [34] is the most influential research model that is used to explain, determine, and understand the recent technology innovation adoption behavior of an individual. Usefulness, ease of use, and behavioral intention are used to assess and evaluate the actual adoption of recent technology, where ease of use and usefulness are the most important TAM determinants that predict individuals' behavior toward acceptance of new technology [34]. Davis et al. [34] stated that perceived ease of use and perceived usefulness have a direct influence on a technology's prospective usage intention, where the degree to which individuals are interested in utilizing the system is characterized as an attitude that produces behavioral intentions and leads to real system adoption and utilization.
Davis et al. [34] defined perceived usefulness as the "subjective perception of users where they believe that using certain technologies can improve the performance of their work." Previous research [35][36][37] has found that people will be encouraged and motivated to adopt and use smartphone technologies if they believe they will be more useful and productive. Davis et al. [34] argue that perceived usefulness and perceived ease of use are crucial factors that influence the intention to adopt and use technology. The studies [22,32,38] investigated the relationship between perceived usefulness, perceived ease of use, and the behavioral intention to adopt and use mobile technologies. Alalwan et al. [32] and Shaw and Kesharwani [22] found that perceived ease of use has no influence on behavioral intention to use smartphones, while [38] reported that perceived ease of use has a positive influence on behavioral intention to use smartphone services. On the other hand, [22,32,38] found that perceived usefulness positively influences individuals' intention to adopt smartphone technologies. Accordingly, the following hypotheses are proposed: Hypothesis 1. Perceived Usefulness (PU) has a positive influence related to children's behavioral intention (BI) to use smartphones.

Hypothesis 2.
Perceived ease of use (PEoU) has a positive influence related to children's behavioral intention (BI) to use smartphones.
The study [31] suggested trust as a critical factor that should be integrated into the TAM model. Also, [30] suggested an extended model based on TAM along with social influence and trust to evaluate the adoption of mobile internet commerce by individuals. Morgan and Hunt [39] and De Wulf et al. [40] defined trust as "consumer confidence in a retailer's reliability and integrity." Tiwari et al. [41] stated that trust is a positive perception of consistency and reliability. According to statistics, 50% of smartphone owners claimed that they do not adopt smartphone financial services due to trust and privacy issues [42,43]. Previous research [22,32] discovered that trust has a considerable influence on intention behavior to adopt and use a smartphone and mobile technologies, leading to the following hypothesis: Previous studies stated that enjoyment motivation is considered an important factor in terms of individuals' intention to adapt and use technologies, systems, and applications [36,[44][45][46]. Alalwan et al. [47], Chong [48], and Dai and Palvi [49] stated that using mobile technology could provide individuals with a degree of enjoyment and fun. [32,36,50] studied the effect of perceived enjoyment as a significant predictor of an individual's behavioral intention to use smartphones. They found that perceived enjoyment influences an individual's behavioral intention to adopt and use smartphone services, applications, and games. Thus, the following hypothesis is Venkatesh et al. [51] defined social influence as "the extent to which an individual perceives that important others believe he or she should apply the new system." According to [52,53], social influence is the individual's decision to accept and utilize modern technology that might be influenced and inspired by friends, close relatives, significant persons they respect, and others in the community. Baabdullah [54] stated that social influence plays a vital role in driving individuals' behavioral intentions toward adopting and using smartphones and mobile applications. Consequently, Baabdullah [54] and Shaw and Kesharwani [22] found that social influence has a positive impact on the behavioral intention of using smartphone applications and games. In this study, social influence is used to evaluate other people's impact on a family's children's adoption and use of the smartphone. Accordingly, the following hypothesis is proposed: Hypothesis 5. Social influence (SI) has a positive influence on children's behavioral intention (BI) to use smartphones.
Individuals' behavioral intentions, according to studies [54][55][56][57], are the desire and readiness to conduct and perform actions that will result in productive outcomes as planned and predicted. According to [51], behavioral intention is "the dependent factor which assesses the behaviors of the individuals towards the used technological service", where individuals are influenced by behavioral intention to make a reasonable effort and to participate in activities that will provide them with the intended services and advantages [32,37,58]. Researchers [22,37] investigated the association between behavioral intention to use smartphones and smartphone addiction. Also, other researchers [2] studied the prevalence of mobile phone addiction among children and teenagers; [2] claimed that smartphone addiction among children and teenagers needs immediate and urgent attention. Shaw and Kesharwani [2,22] found that children's smartphone addiction is influenced positively by behavioral intention. Thus, the following hypothesis is suggested: Hypothesis 6. Behavioral intention (BI) has a positive influence relating to children's smartphone addiction (SPA).
Technology addiction is "a special type of behavioral addiction that encapsulates a psychological dependency on the use of an IT", according to Turel et al. [59]. Smartphone addiction is "an emerging concept in which consumers maximize their smartphone usage in various activities and exhibit changes in behavior," according to Shaw and Kesharwani [22]. Meshi and Ellithorpe [29] investigated the connections between social media use and social isolation, depression, and anxiety. According to [2], smartphone addiction in kids and teenagers has become a widespread issue that affects everyone. Furthermore, [2] argued that while internet addiction has received most of the study attention, a thorough and in-depth examination of smartphone addiction has lagged. Additionally, [19] discovered that children and teenagers are more attracted to smartphones and show more signs and symptoms of addictive behavior. According to [25], technology has a significant impact on young people's behavior, and smartphone addiction and social isolation (SocIso) are positively correlated. The social isolation (SocIso) items in this study were measured using five items that were adopted from the Patient-Reported Outcomes Measurement Information System (PROMIS) [60]. As a result, the following assertion is made: Hypothesis 7. Smartphone addiction (SPA) has a positive influence on children's social isolation (SocIso).
Previous studies studied the student's smartphone use addiction in terms of different variables, including the parents' age, gender, educational level, school environment, and other variables. The studies [61,62] argue that another factor that might influence the children's screen time problem might be the parent's gender [61,62], and researchers [61,63,64] have investigated the relationship between the parents' age and the children's smartphone In Jordan, children attend nurseries to get childcare from the ages of a few weeks to 3 years. Later, from 4-5 years old, all children are required by Jordanian law to attend preschool to learn and gain the basic skills required to attend compulsory education in schools from 6-18 years old. There are three types of schools in Jordan: governmental, UNRWA, and private schools. On the one hand, the governmental and UNRWA schools provide a national program developed by the Ministry of Education (MOE) that is taught in the Arabic language [71]. On the other hand, private schools provide either national or international programs where the international programs are offered and taught in the English language, such as International GCSE, the International Baccalaureate (IB), and the Scholastic Assessment Test (SAT) [72,73]. Research [5] found that students enrolled in English-based educational programs spend more time on smartphones than students enrolled in French-based educational programs, with time spent by students in the English programs exceeding two hours per day compared to children enrolled in French programs. Also, [65] found that the school environment increases the school students' problematic smartphone use. Therefore, the following hypothesis is proposed: Hypothesis 15. A child's school has a positive influence on the child's smartphone addiction (SPA).
Studies report that children's daily screen time is rapidly increasing and begins at an early age (5)(6)(7)(8). Teresia et al. [16] stated that time spent using smartphones is the "frequency and duration that youth are engaged in screen media use." Christakis and Zimmerman [6] stated that the percentage of newborns that watch TV on a regular basis increased dramatically from 40% to 90% by the age of two years. Study [62] reported that American children and teens spend an average of more than seven hours per day on screen media. As a result, the following hypothesis is proposed to address the relationship between the amount of time children and teenagers spend using smartphones and smartphone addiction: Hypothesis 16. Time spent using smartphones has a positive influence on children's smartphone addiction (SPA).
Smartphones provide the ability to access the internet in addition to a wide range of apps such as messaging, gaming, and social networking. Numerous investigations have been conducted into the elements that affect smartphone addiction, including the installed apps and games [19,50,74,75]. Research [50] found that the high quality of games' interfaces significantly influences smartphone addiction. Study [74] stated that perceived enjoyment of a smartphone game is positively associated with smartphone addiction. Study [75] found that overuse and the lack of control over the time spent on SNS apps are the main causes of the social networking apps' addiction. Accordingly, we suggest the following hypothesis: Hypothesis 17. Favorite smartphone app/game/SNS has a positive influence relating to children's smartphone addiction (SPA).

Research Methods
For this study to achieve its purpose of investigating and examining the overall effect of smartphones on children's and adolescents' social isolation (SocIso), it examines the impact of the independent factors perceived usefulness (PU), perceived ease of use (PEoU), trust (TR), perceived enjoyment (EN), and social influence (SI) on the intermediate variable, behavioral intention; the effect of the intermediate variable, behavioral intention, on smartphone addiction (SPA); and the effect of smartphone addiction (SPA) on social isolation (SocIso). Furthermore, the researchers studied the moderating roles of parents' age, education level, gender, and marital status, in addition to the number of children in the family. Also, this study investigated the moderating role of children's gender, age, school, and the time that the child spent using a smartphone, as well as their favorite smartphone app, game, or social network (SNS).
As previous investigations into this topic have been limited or incomplete, the researchers, after an extensive research study and development stage, suggested the research model introduced in Figure 1, in addition to the proposed hypotheses. Moreover, a questionnaire was used and evaluated, and data from 511 participants was collected from a convenience sample. The following three parts are presented to clarify and explain in detail the survey design and methodologies of this research.

Research Context
Smartphone addiction and the problem of social isolation among children and adolescents have emerged as serious problems in society. As a result, the key questions are: what are the determinants that influence children's smartphone addiction, and is there a significant correlation between social isolation and smartphone addiction? This study was carried out as follows.

Measurement Items
A questionnaire survey was developed to evaluate the proposed research model. Previous studies were used to develop the survey items and the study model contains eight direct and intermediate factors as well as ten moderating factors.
Perceived usefulness (PU) and perceived ease of use (PEoU) were adopted from [22,32]. Perceived usefulness (PU) was measured by six items, while perceived ease of use (PEoU) was measured by four items. Trust (TR) was measured by five items according to [22,32]. Perceived enjoyment (EN) was measured by three items adapted from [32,50]. Social influence (SI) was measured by four items adopted from [22,54]. Behavioral intention (BI) was measured by three items adapted from [2,22]. Smartphones addiction (SPA) was measured by nine items adopted from [19,25]. Social isolation (SocIso) was measured by five items adapted from (PROMIS) [60]. Constructs and items are reflected in detail in Table A1 in Appendix A.

Participants and Procedures
A web-based Google form was developed to collect data. The objective was to validate and examine the research model and hypotheses using a 5-point Likert scale: (1) strongly disagree; (2) disagree; (3) neither agree nor disagree; (4) agree; and (5) agree strongly. The constructs and items used to measure the constructions, as well as the mediating variables, are summarized in Table A1 in Appendix A.
The survey was developed in both Arabic and English and a panel of eight academics examined it; accordingly, the questionnaire was revised in response to their feedback. Also, the questionnaire was piloted on 20 Jordanian parents to ensure that the questions were understandable; consequently, the survey has been revised. The target population of this study involved all children and adolescents who use smartphones in Jordan, where the study data was collected from their parents who responded through the convenient sampling technique from 21 May 2022 to 12 June 2022. The survey was distributed through e-groups including Facebook and WhatsApp. Table 1 shows the responses from the respondents' parents.
Although the respondents were 523 in number, 12 had to be eliminated since those respondents did not complete the questionnaire. Hence, the demography of this study consisted of 511 parents the sample size is used according to Morgan table (recommending 384 responses), when sample size is unknown as in this case. Further, the researchers checked for duplication using a Microsoft Access query for duplication (Find duplicate query) on all fields. As can be seen in Table 1, the demographic profile of the respondents was male (72%) with an age of 28-48 years (84%). The profile also includes the education of most as B.Sc. (52%), and they were married (97%). The questions were answered on behalf of the children, with the ages 3-6 years old (32%) and 6-9 years old (20%). Hence, 52% of the children are ages 3-9 years old. The people who answered the survey (84%) think that the kids' internet experience is either excellent or good. Also, (62%) of the families had 2-3 children. They described the use of the internet in hours as (23%) of them used the internet for more than 5 h, while others used the internet for 2 h or 3 h (21%), for a total of (42%). The children study in government/public schools (69%) and follow YouTube Kids (59%) and use games (25%).

Data Analysis and Results
The analysis of data for this study included: first, a descriptive analysis to measure respondents' attitudes; second, a structural equation model (SEM) (which included a confirmatory factor analysis (CFA) and then structural equation modeling (SEM) using Amos 26, performed to test the study hypotheses); third, the moderating effects; and, finally, validation of this research using machine learning (ML). SEM and CFA verified the hypotheses and analyzed the results whilst ML validated and predicted mean square error and correlation coefficient (R 2 ). This research employed triangulation by using multiple data collection and analysis.

Descriptive Analysis
The mean and standard deviation were estimated to describe the responses and thus the attitude of the respondents toward each question asked in the survey. While the mean represents the data's central tendency, the standard deviation measures its dispersion and provides an index of the data's spread or variability [76,77]. A small standard deviation for a set of values indicates that these values are clustered closely around or close to the mean; a large standard deviation indicates the opposite. The level of each item was determined by the following: (1) Hence, producing the following lookup Table 2 of values:  Table 3 shows the constructs with mean, standard deviation (SD), level, and order. All constructs were ranked "High" to "Very High" according to Table 2 based on the work of [76,77]. The exception is the construct TR, which ranks "Low" with a mean below (3). The construct EN ranked as the first among all. Both mediating constructs were ranked "High" as was the dependent construct, SocIso.  Table 4 presents the mean, standard deviation, level, and order of the constructs with the items in addition to Cronbach Alpha for each construct. Cronbach's Alpha is a measure for reliability and consistency in multiple-question Likert scale surveys. The range is expected to be greater than 0.7, while anything less than 0.70 is considered low. Cronbach's Alpha above 0.9 is considered excellent internal consistency, greater than 0.8 is considered good internal consistency, while between 0.7 and 0.8 is considered acceptable. On the other hand, below 0.7 is considered questionable internal consistency. As can be seen from Table 4, all constructs are reliable with Cronbach Alpha above 0.70 except BI. In the later stage of the study, the authors had to withdraw BI2 since it was the source of the discrepancy, and the Cronbach Alpha improved to become (0.75266775). Later, we determined that the question in Arabic was vague.

SEM Analysis
In this section, a measurement model assessment was conducted, as were model fit assessment and model reliability and validity and structural model assessment.

Measurement Model Assessment
CFA was used to evaluate the properties of the instrument items. In fact, the measurement model signifies how hypothetical constructs are measured in terms of the observed variables and personifies the validity and reliability of the observed variables' responses to the latent variables as in [78][79][80]. Table 5 presents the factor loadings, composite reliability (CR), and average variance extracted (AVE) for the variables. All the indicators of the factor loadings exceeded 0.50, except for certain items, specifically BI2. The aforementioned items were eliminated, thus constituting evidence of convergent validity as in [78,81]. While the measurement reached convergent validity at the item level because all the factor loadings were above 0.50, all the composite reliability (CR) values exceeded 0.60, demonstrating an important level of internal consistency for the latent variables. Additionally, since each value of AVE exceeded 0.50, as in [78,79] the convergent validity was proved.

Model Fit Assessment
The proposed model's fitness was evaluated using the following fit indices. As can be seen in Table 6, the model passed all the recommended tests. The chi-square due to the sample size is p = 0.000, CMIN = 1184, DF = 629. Furthermore, CMIN/DF is the discrepancy divided by the degree of freedom, which in this study is less than 5.0 and less than 3.0, as recommended by [82]. If the CMIN/DF value is ≤3 it indicates an acceptable fit [80]. The baseline comparisons are CFI, IFI, and NFI. CFI is the Comparative Fit Index and has a value truncated between 0 and 1, where values close to 1 show a very good fit while 1 represents the perfect fit [83]. The value of interest here is CFI for the default model. A CFI value of ≥0.95 is considered an excellent fit for the model [84]. An Incremental Fit Index (IFI) where values are close to 1 indicates a very good fit, while 1 indicates a perfect fit. In this study, IFI = 0.963. For the Normed Fit Index (NFI), also referred to as Delta 1, a value of 1 shows a perfect fit, while models valued < 0.9 can usually be improved substantially [85]. In this study, NFI = 0.925. The Parsimony-Adjusted Measures are PCFI and PNFI. PNFI is the Parsimony Normed Fixed Index, expressing the result of parsimony adjustment [86] to the Normed Fixed Index (NFI). In this study, PNFI was 0.842, which is greater than 0.5 according to [87]. PCFI is the Parsimony Comparative Fix Index, expressing the result of parsimony adjustment applied to the Comparative Fit Index (CFI). In this study, PCFI was 0.876, which is greater than 0.5 according to [87]. RMSEA stands for Root Mean Square Error of Approximation, and values greater than 0.1 are considered poor, values between 0.08 and 0.1 are borderline, values between 0.05 and 0.08 are acceptable, and values less than 0.05 are considered excellent [88], as is the case here with 0.042. According to [79], when SRMR is less than or equal to 0.09, it indicates an acceptable fit, and in this study, SRMR = 0.0629. To validate the proposed model, first construct reliability was conducted and then convergent validity. Four indicators were calculated using AMOS 26, CR, AVE, MSV, and MaxR(H). AMOS 26 suggested the removal of BI2 since AVE = 0.399 and it should be greater than 0.5. Hence, after the removal of item BI2, the analysis was conducted again. Because CR and MaxR (H) are both greater than 0.7, construct reliability has been established. Convergent validity is established when AVE used for convergent validity is greater than 0.5. MaxR(H) (Maximal Reliability) is >0.7. The values of CR, AVE, and MaxR(H) in Table 11 show that the model is reliable and valid. Results are shown in Table 7, and the AMOS 26 indicated that there is no validity concern in the results. No validity concerns here, according to [83] using [76]. Table 8 shows the correlation among construct thresholds based on [83] using [76]. The diagonal elements in the table are the square root of AVE, and all correlations between constructs are less than the square root of AVE, indicating that they are all statistically significant. One may note here that there is strong correlation between SocIso and SPA, and moderate correlation between BI and both PU and TR. Significance of Correlations: * p < 0.050, ** p < 0.010, *** p < 0.001; thresholds from [83] using [76].
The heterotrait-monotrait ratio of correlations (HTMT) criterion measures the average correlations of the indicators across constructs. The acceptable levels of discriminant validity are (<0.90), as suggested by [79] and developed by [92]. Table 9 below reflects the results.

Structural Model Assessment
As stated previously, the model is fit. Next, we will discuss estimating the path coefficient (hypothesis testing) and estimating squared multiple correlation R 2 . Figure 2 shows the path coefficient using AMOS 26. The figure shows the R 2 highlighted above the intermediate constructs and dependent constructs, BI, SPA, and SocIso (0.70, 0.10, and 0.52, respectively). Further, Table 10 reflects the coefficients where two hypotheses were not supported, namely H2 and H4, since p-Value> 0.05. On the other hand, the findings reflect in the same table that H1, H3, H5, H6, and H7 are all supported by the findings of the study.

Moderation Effects
The study investigated the significance of two bi-variable groups, the gender of both the child and the respondent, and their effects on SPA, BI, and SocIso, and found that both variables had no significance, as per the group statistic Tables 11 and 12.  The outcomes of the ANOVA test, presented in Table 13, indicate the following: there is a significant difference in the respondents' BI, supportive of the respondent's age, supportive of internet experience, and child age. There is a significant difference in the respondents' SocIso supportive of both the number of children in the family and the hours spent on a smartphone. There is a significant difference in both the respondents' BI and SPA, which is supportive of the number of hours spent on smartphones. Table 14 provides the statistical significance of the differences between each pair of groups for respondents' age. As shown in Table 13, the five groupings were statistically different from one another.  Table 15 provides the statistical significance of the differences between each pair of groups for children's age. As observed in Table 14, the six groupings were statistically different from one another.  Table 16 provides the statistical significance of the differences between each pair of groups for internet experience. As observed in Table 15, the three groupings were statistically different from one another.  Table 17 provides the statistical significance of the differences between each pair of groups for the number of children in a family. As observed in Table 16, the five groupings were statistically different from one another.  Table 18 provides the statistical significance of the differences between each pair of groups for the number of hours spent on the smartphone. As observed in Table 17, the five groupings were statistically different from one another. Table 19 provides the statistical significance of the differences between each pair of groups in hours on the phone. As observed in Table 19, the five groupings were statistically different from one another.   Table 20 provides the statistical significance of the differences between each pair of groups' hours on the phone. As observed in Table 20, the five groupings were statistically different from one another.

Machine Learning Techniques Validation and Prediction
Machine learning techniques have been used as modern technologies in different applications [93,94]. Further, other studies like [95][96][97][98][99][100][101] used such methods for triangulation method to validate and verify the results along with SEM. The research [102] used 19 machine learning techniques. Five Machine Learning (ML) classification techniques are evaluated in this study, which transform inherited data from a dataset's input into the required output pattern [94,103]. The five ML models used to develop and evaluate models for the smartphone isolation dataset application are: Artificial Neural Network (ANN) [104], Linear Regression [105], Sequential Minimal Optimization algorithm (SMO) for Support Vector Machine (SVM) [106], Bagging using REPTree model [107], and Random Forest [108]. The ANN employs the back-propagation method to calculate the errors between the predicted and actual output values. The weights and bias parameters of the ANN design are then modified using the error to bring the predicted and actual values to be closer. The output of the linear regression model depends on the target labels and is a polynomial function with weighted coefficients for the independent variables. Through a sequence of actions, the training phase updates the linear function's coefficients from the training dataset. The SMO method updates the weighted vectors of the SVM model using the Sequential Minimal Optimization algorithm. The SMO algorithm finds the minimal values in a sequence of iterative operations to reach the optimal values. The bagging technique constructs numerous REPTree models using a random sample of the training set's instances and features, with the average value of the trees predicting the final value. The Random Forest (RF) is a set of connected decision tree (DT) models built by random attribute subsets for each sub-tree model and a random sampling of training data instances. The average value of the DT trees serves as the model's final output.
The evaluation methodology follows 10-fold cross-validation technique to validate the effectiveness of the model to predict the target values. During the evaluation phase, the 10-fold cross-validation method is used. This method sequentially selects 10% of the dataset as testing and 90% as training (the remaining nine folds). We create a classifier model and assess how well it performs in each procedure. Then, a representation of the overall average performance is shown. By employing such a method, we ensure that the complete dataset is used during the training and testing stages, lowering the possibility of over-fitting. When the model successfully categorizes all the training data but is unable to fit the test sets, a problem arises.

ML Results and Discussion
Children who use smartphones for extended periods face two serious problems: smartphone addiction and social isolation. This study investigates a few aspects that influence the two problems and validates certain integration techniques. To understand the relationship between the factors (or inputs) and the problems, ML techniques as intelligent methods extract inherited meaningful information from datasets. However, to assess the performance of ML models, we need three datasets. The datasets are from model 1, which has BI as a dependent outcome and five parameters (PU, PEU, Trust, PE, and SI) as independent inputs. Model 2 dataset studies the influence of BI as input to SOA as a dependent variable. Model 3 of the dataset represents the impact of SOA on SI. Figure 3 displays the experimental findings utilizing R 2 and Mean Square Error (MSE) as evaluation metrics. The models are shown on the x-axis, and the R 2 and MSE values are shown on the y-axis. The R 2 shows the expected impact of the independent variables on the dependent variable (target). The MSE calculates, as in Figure 4, the average discrepancy between the predicted and actual output values of a model. g Data Cogn. Comput. 2022, 6, x FOR PEER REVIEW Children who use smartphones for extended periods face two seriou smartphone addiction and social isolation. This study investigates a few as fluence the two problems and validates certain integration techniques. To un relationship between the factors (or inputs) and the problems, ML techniqu gent methods extract inherited meaningful information from datasets. Howe the performance of ML models, we need three datasets. The datasets are fr which has BI as a dependent outcome and five parameters (PU, PEU, Trust, P independent inputs. Model 2 dataset studies the influence of BI as input to pendent variable. Model 3 of the dataset represents the impact of SOA on SI Figure 3 displays the experimental findings utilizing R 2 and Mean (MSE) as evaluation metrics. The models are shown on the x-axis, and the values are shown on the y-axis. The R 2 shows the expected impact of the variables on the dependent variable (target). The MSE calculates, as in Figur age discrepancy between the predicted and actual output values of a model.
On three database models, the SMO and linear regression sequential mo reasonable results, with R 2 values of 65%, 20%, and 68%, respectively. The ot niques that are non-linear methods, such as ANN, Bagging REPTree, and Ra obtain convergent results. The results show a weak relationship between the smartphone addiction of 20% R 2 and 95% MSE in model 2, which indicates t tion of using smartphone by children provides less information to detect the model 1, the five factors reflect how usability of smartphones affects the tend its applications by the children and their parents with 65% R 2 value and ap 49% MSE value. Moreover, with the existence of smartphone addiction, so inevitably occurred, which is indicated by the R 2 value of 68% in model 3. T niques as validation techniques are able to predict the actual target from the inputs and ensure the provided results.

Discussion
As stated in the introduction, the purpose of this study is to provide a better un standing and insight into the primary factors that impact children's smartphone ad tions (SPA), as well as the relationship between smartphone addiction (SPA) and so isolation (SocIso). The main empirical findings of this study show that most of the posed model's hypotheses have significant values. Consequently, the current study's f ings are consistent with previous literature and this study yielded several notable con sions and findings.
As predicted, perceived usefulness was found to have a significant influence on danian children's and adolescents' behavioral intentions to adopt and use smartpho This may be related to the advantages and conveniences that children and adolesc gain from using smartphones, as stated by references [22,32,[35][36][37] that individuals wo be encouraged to use smartphones as long as they consider such technologies more us and productive. On the other hand, perceived ease of use is an important factor tha fluences the intention to adopt and use smartphone technology [38]. The results of study, from the perspective of parents, show that perceived ease of use has no influe on Jordanian children and adolescents' behavioral intention to adopt and us smartphone, which is in line with what has been found by studies [22,32,37].
Furthermore, the findings show that trust has a positive influence on the childr and adolescents' behavioral intentions to use smartphones, which is consistent with previous results of the studies [22,32]. This means that parents have a positive percep of the consistency and reliability of smartphone applications, and they allow their child to use the smartphone freely. On the other hand, the findings of this study related to trust variable are not compatible with the results of Tiwari et al. [41] as they stated trust and privacy issues are the main reasons that more than 50% of smartphone ow claimed they do not adopt and use smartphone financial services.
While enjoyment motivation is considered an important factor in terms of indiv als' intention to adapt and use technologies [36,[44][45][46], the study results show that ceived enjoyment has no influence related to children's and adolescents' behavioral in tion to use a smartphone. Such results are inconsistent with the findings of the stu [32,36,50] which stated that perceived enjoyment influences individuals' behavioral in tion to adopt and use smartphone services, applications, and games. This means that e when children or adolescents feel that using the smartphone is not enjoyable, they keep using it, according to how parents responded to the questionnaire. On three database models, the SMO and linear regression sequential models produce reasonable results, with R 2 values of 65%, 20%, and 68%, respectively. The other ML techniques that are non-linear methods, such as ANN, Bagging REPTree, and Random Forest, obtain convergent results. The results show a weak relationship between the BI factor and smartphone addiction of 20% R 2 and 95% MSE in model 2, which indicates that the intention of using smartphone by children provides less information to detect the addiction. In model 1, the five factors reflect how usability of smartphones affects the tendency of using its applications by the children and their parents with 65% R 2 value and approximately 49% MSE value. Moreover, with the existence of smartphone addiction, social isolation inevitably occurred, which is indicated by the R 2 value of 68% in model 3. The ML techniques as validation techniques are able to predict the actual target from the independent inputs and ensure the provided results.

Discussion
As stated in the introduction, the purpose of this study is to provide a better understanding and insight into the primary factors that impact children's smartphone addictions (SPA), as well as the relationship between smartphone addiction (SPA) and social isolation (SocIso). The main empirical findings of this study show that most of the proposed model's hypotheses have significant values. Consequently, the current study's findings are consistent with previous literature and this study yielded several notable conclusions and findings.
As predicted, perceived usefulness was found to have a significant influence on Jordanian children's and adolescents' behavioral intentions to adopt and use smartphones. This may be related to the advantages and conveniences that children and adolescents gain from using smartphones, as stated by references [22,32,[35][36][37] that individuals would be encouraged to use smartphones as long as they consider such technologies more useful and productive. On the other hand, perceived ease of use is an important factor that influences the intention to adopt and use smartphone technology [38]. The results of this study, from the perspective of parents, show that perceived ease of use has no influence on Jordanian children and adolescents' behavioral intention to adopt and use a smartphone, which is in line with what has been found by studies [22,32,37].
Furthermore, the findings show that trust has a positive influence on the children's and adolescents' behavioral intentions to use smartphones, which is consistent with the previous results of the studies [22,32]. This means that parents have a positive perception of the consistency and reliability of smartphone applications, and they allow their children to use the smartphone freely. On the other hand, the findings of this study related to the trust variable are not compatible with the results of Tiwari et al. [41] as they stated that trust and privacy issues are the main reasons that more than 50% of smartphone owners claimed they do not adopt and use smartphone financial services.
While enjoyment motivation is considered an important factor in terms of individuals' intention to adapt and use technologies [36,[44][45][46], the study results show that perceived enjoyment has no influence related to children's and adolescents' behavioral intention to use a smartphone. Such results are inconsistent with the findings of the studies [32,36,50] which stated that perceived enjoyment influences individuals' behavioral intention to adopt and use smartphone services, applications, and games. This means that even when children or adolescents feel that using the smartphone is not enjoyable, they will keep using it, according to how parents responded to the questionnaire.
As stated previously in this study, social influence is used to evaluate other people's impacts on adopting and using the smartphone. The results show that social influence has a positive influence on the children's behavioral intention to use smartphones, which confirms the findings of Baabdullah [54] and Shaw and Kesharwani [22] that showed the positive impact of social influence on the behavioral intention towards using smartphone applications and games.
Also, this study investigated the association between behavioral intention to use smartphones and smartphone addiction among Jordanian children and teenagers. The findings show that there is a link between behavioral intention and smartphone addiction, with behavioral intention having a positive influence on the smartphone addiction of children and adolescents. Such findings are in line with the results provided by Shaw and Kesharwani [2,22] which reported that smartphone addiction among children and teenagers needs immediate and urgent attention. Parents responded to the survey by saying that they would not recommend others to let their children use smartphones, as reported in the (BI4) item.
And finally, the main empirical result of this study investigates and detects the association between smartphone addiction and social isolation using items adapted from the Patient-Reported Outcomes Measurement Information System (PROMIS) [60], where most of the previous studies have been focused on internet addiction, but a comprehensive and detailed review of smartphone addiction is inadequate till now, even though smartphone addiction among children and teenagers has become a public and major problem for everyone [2,19,22,29]. The study findings prove that there is a positive association between smartphone addiction and social isolation among children and adolescents in Jordan, which is consistent with the results and findings of [25].
Also, the demography of this study shows that the respondents were male (72%). On the other hand, in the study proposed by Anderson et al. [62], researchers stated that most of the respondents were mothers. Also, we think that we must mention that we received a lot of messages from male parents (fathers) asking how they can solve the problem of smartphone addiction and social isolation, in addition to other messages asking about the results of the study and if the problem is a major problem in society. The study found that the gender of both child and respondent had no significance either in behavioral intention or in smartphone addiction and social isolation of children and adolescents in Jordan. The findings also show that there is a significant difference in the respondents' behavioral intentions supportive of the respondent's age, supportive of internet experience, and child age. Such results are in line with [61,63,64]. Also, there is a significant difference in the respondents' social isolation supportive of both the number of children in the family and hours spent on smartphones, which is consistent with the findings of [62,67]. Moreover, there is a significant difference in both the respondents' behavioral intention and smartphone addiction, which is supportive of the number of hours spent on smartphones, which is consistent with the results of [62].

Theoretical Implications
The purpose of this study was to investigate various independent, mediating, and moderating determinants that influence smartphone addiction and social isolation, and to detect the association between smartphone use, addiction, and social isolation among Jordanian children and teens. Many studies have been carried out to have a further and better understanding and knowledge of this critical issue, as mentioned in the introduction and the research hypothesis development sections, even though no research has been identified that incorporates all these factors into a single study of smartphone addiction and social isolation. Also, to date, attention has been focused on internet addiction, but a detailed investigation of smartphone addiction and the correlation between addiction and social isolation is lacking [2,22,25,29].
This study will add to the literature and the body of knowledge regarding the correlation between smartphone use, smartphone addiction, and social isolation. The study detected and discovered the effects of five independent components, with ten potential moderators on behavioral intention as intermediate factors, as well as smartphone addiction and social isolation. Moreover, the research included many moderating factors that are unique to such research, which are the children's internet experience, time spent using smartphones, the number of children in the family, and parent education level.

Practical Implications
Children and teens are the primary users of smartphones, and their daily screen time is rapidly increasing at an early age, even with many recommendations to control and set a time for children and adolescents to use smartphones, since children and adolescents should use these devices as little as possible. Furthermore, studies have revealed that children and teenagers are more drawn to smartphones, as well as exhibiting more addictive behavioral signs and symptoms with them. Such technology highly affects youth behavior, and there is a positive association between smartphone addiction and social isolation.
Accordingly, smartphone addiction and social isolation among children and teenagers require immediate and urgent attention. The government should develop awareness programs to educate parents and children about the dangers of using smartphones and the related issues associated with smartphone addiction and social isolation. Moreover, the study findings show there is a significant difference in the parents' behavioral intention supportive of the parents' age, where parental control over children's smartphone usage decreases as the age of the parent increases. Hence, the government should enable older parents by training and education about smartphone control programs and how to set a time limit and determine the applications and games that the children can install and use on their smartphones. Furthermore, governments are responsible for developing social activities, camps, and other physical activities for children and adolescents.
On the other hand, parents must encourage their children and teenagers to join in such activities and socialize with their peers. Moreover, parents must not recommend others let their children use smartphones because of the addictive behavioral signs and symptoms associated with smartphone use, whereas real-life social interaction reduces social isolation.
The study's findings could be utilized as resources for early diagnosis and detection of children and teens at risk of smartphone addiction and social isolation. In addition, they could be utilized to develop prevention programs to reduce smartphone overuse among children and teenagers. Also, the study findings provide valuable information. This can help and support a campaign against smartphone addiction, social isolation, and overuse.

Academic Implications
Based on the study's findings, it contributes significantly to the literature investigating and detecting smartphone addiction and social isolation. The study model was developed using the TAM model, with the addition of trust, social influence, and perceived enjoyment. Furthermore, the findings show that the study model is robust and significantly interpretable, which contributes to the studies of smartphone adoption, addiction, and social isolation. Also, according to the proposed model and the findings of this study, researchers effectively presented and demonstrated a fundamental association between the impact of the determinants and technology addiction and social isolation. This significantly extends TAM's theoretical purview to be used in future studies to improve the investigation of smartphone adoption, usage, technology addiction, and social isolation.

Limitations and Future Research
In this study, the researchers faced two main limitations. First, because the study was conducted on one of society's most pressing issues, smartphone addiction and social isolation among children and teenagers, it was difficult to gain access to participants, even though several channels were used to increase the number of participants, such as WhatsApp groups, Facebook, etc. The researchers only gathered 511 responses. Another limitation was the underrepresentation in some of the demography categories: in marital status (divorced and widow) were 2% and 1% of the respondents. Respondent age groups (18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28) were 3%. Preferred SNS did not find much interest in LinkedIn and Twitter, since both intuitively relate more to adults rather than children.
Future research should investigate whether artificial intelligence (AI) ML methods can be used to investigate, predict, detect, and prevent children's intentions to engage in smartphone addiction or social isolation behavior.
In this study the researchers tried five different methods: Artificial Neural Network (ANN), Linear Regression, Sequential Minimal Optimization algorithm (SMO) for Support Vector Machine (SVM), Bagging using REPTree model, and Random Forest, which distinguish this work. There are other ML methods that can be experimented with, such as k-NN, ID3, and Naïve Bayes.
Also, it would be worthwhile to study the impact of smartphone usage and addiction on children's health and physical ability. Another suggestion for future work is to study the relationship between parents' smartphone addiction and their children's addiction. Furthermore, future studies must consider the smartphone overuse prevention models and study the effects of other smart devices and the relationships between such devices and technology addiction and social isolation. Also, they could study the association between smartphone addiction and children's academic achievements. Finally, the proposed model can be expanded to include other constructs.

Conclusions
This research investigates the determinants that influence children's smartphone addiction and the association between social isolation and smartphone addiction for people living in Jordan. The proposed model is developed using the original TAM model in addition to social influence, trust, and perceived enjoyment constructs. The collected study data were examined using structural equation modeling (SEM), machine learning (ML), and computational fluid dynamics (CFA). According to the respondents' responses, the results showed that perceived usefulness, trust, and social influence were significant antecedents to behavioral intention to use the smartphone. Furthermore, the findings confirm that smartphone addiction positively influences social isolation among Jordanian children and adolescents, where the strength of these correlations is influenced by moderating variables, including respondent's age, child internet experience, and child age, as well as the number of children in the family and hours spent on smartphones. On the other hand, perceived ease of use and perceived enjoyment did not have a significant effect on behavioral intention to use the smartphone among Jordanian children and adolescents.
Therefore, we conclude with the following intriguing conclusion: First, the results showed that parents think of smartphones as devices that can be useful and can be trusted for use by their children, and which have been adopted and used according to others' recommendations, while the study shows that most of the children mainly follow YouTube for Kids (59%) and use games (25%), and that many (23%) of them use the internet more than 5 h per day. As mentioned in the literature, children who use smartphones for a long time exhibit addictive behavioral signs and other symptoms such as social isolation, which require immediate and urgent attention and suitable action and early treatment.
Secondly, according to the study findings, parental control over children's smartphone usage decreased as the age of the parent increased. Hence, older parents should be enabled by training and education about smartphone control programs and how to set a time limit and determine the applications and games that children can install and use on their smartphones. Third, parents must encourage their children to join different activities and socialize with their peers. Finally, this study contributes to the literature by empirically examining and theorizing the implications of smartphone addiction on social isolation.  Behavioral Intention (BI) BI1: I intend to let my child use the mobile phone in the future. BI2: My child is using the smartphone, and he/she always tries to use it whenever he/she can at any time. BI3: I plan to keep my child's smartphone in use in the future. BI4: I will recommend that others let their children use smartphones.
Smartphone Addiction (SPA) SPA1: My child sometimes ignores important things because of his/her interest in smartphones. SPA2: My child often fails to get enough rest because of using a smartphone. SPA3: My child's social life has sometimes suffered because of using a smartphone. SPA4: Arguments have sometimes arisen from people around me because of the time my child spends on smartphones. SPA5: Using a smartphone has sometimes interfered with my child's studying, playing, or social activities. SPA6: My child is sometimes late for engagements (like studying) because of using smartphones. SPA7: When my child is not using a smartphone, I feel that he/she often feels agitated and confused. SPA8: I have made unsuccessful attempts to reduce the time my child uses a smartphone. SPA9: I think that my child is addicted to smartphones.
Social Isolation (SocIso) SocIso1: I feel that even when children are around my child, they ignore him because he is busy using his smartphone. SocIso2: I feel that other children avoid talking to my child because he is busy using a smartphone. SocIso3: I feel that my child is isolated even if he is with other children because he is busy using his smartphone. SocIso4: I feel that my child is isolated by others because he is busy using a smartphone SocIso5: I feel that my child is isolated from others because he is busy using a smartphone.