Proposing a Method Based on Artificial Neural Network for Predicting Alignment between the Saudi Nursing Workforce and the Gig Framework

Abstract

The goal of the present investigation is to assess the applicability of the Gig Economy Framework (GEF) to the nursing workforce in Saudi Arabia. In order to learn more about the viability of the gig economy paradigm for the nursing profession, this study employed a cross-sectional survey technique. The survey asked questions specific to the nursing profession in Saudi Arabia and the GEF, while also taking into account other relevant variables. This nurse survey was sent to 102 Saudi Arabian hospitals’ HR departments. After removing invalid and missing data, 379 responses remained. The gig economy’s impact on everyday living and professional growth differed significantly between groups. After processing the data, we inputted them into a multi-layer perceptron (MLP) neural network to find relationships between responses to surveys and compatibility with the GEF. There were 20 inputs to this neural network and four possible outputs. The results of the network are the answers to questions about how the gig economy might affect four areas—life, financial management, and personal and professional comfort and development. Outputs 1–4 were predicted with 96.5%, 96.5%, 99.2%, and 99.2% accuracy, respectively. The primary issues with the nursing workforce in Saudi Arabia may be addressed with the use of gig economy elements. As a result, it is crucial to provide a trustworthy, intelligent strategy for foreseeing the gig economy’s framework’s alignment.

1. Introduction

A gig economy is a kind of free market where businesses often use independent contractors for brief assignments, and temporary roles are popular. A task that lasts for a certain amount of time is referred to as a “gig” in colloquial language.

Freelancers, independent contractors, project-based employees, and temporary or part-time personnel are a few examples of gig workers. Customers and gig workers are often connected via digital technologies and gig applications. The gig economy is a recent trend that has risen as a result of several circumstances. The following are the two most crucial factors: The workforce has expanded in mobility, and more and more work is being performed remotely on digital platforms. Jobs and localities are becoming less linked as a consequence. This implies that independent contractors may accept a position or project from an employer anywhere on the globe. The best candidate for a particular project may be chosen by employers from a wider pool than is accessible in any one location. The sharing economy, gift economy, barter economy, and more adaptable employment all form part of the altering cultural and corporate landscape. The gig economy’s cultural effects are still evolving; for instance, employment patterns have been significantly impacted by the COVID-19 epidemic. The flexibility of the gig economy allows people to work freely for particular jobs or tasks supplied by firms without the need for a traditional employment structure or benefits package. The World Economic Forum’s definition of the “gig economy” [1] emphasizes workers’ engagement in the labor market and their ability to earn money via “gigs” or “tasks” for which they are employed. The gig economy’s distinctive employment model, which seems to have numerous advantages over conventional employment, has been fast reshaping the future of labor. There is more freedom of choice for job and working hours, since gig workers are not tied to any employer and may choose to work for anybody who will hire them [2,3,4]. In a similar vein, companies and employers benefit from more mobility, since they are less reliant on a fixed team to complete projects. Because they are not required to provide the same perks, advances, insurance, etc. that are typical of traditional employment, they can also save money. The “gig economy” has recently spread across the entire planet. About a quarter of all US employees perform gig labor of some kind, and one in ten rely on gigs exclusively for income [5]. Forecasts put the total value of the gig economy worldwide at USD 455.2 billion by 2023 [6], up from USD 296.7 billion in 2020. The rise of the freelance worker may be seen across the industrialized world. Eighty-six million Americans are anticipated to work in the “gig economy” by 2027 [7]. This year, there will likely be 4.7 million Britons employed in the Gig economy, up from 2.3 million in 2016 [7]. Sixty percent or more of gig workers say they work gigs at least once a week [7]. About 44% of gig workers in the US regard gig work as their principal revenue source. In a similar line, the economies of developing nations, like India, may be significantly impacted by the gig economy. Ninety million individuals, or over 30% of India’s non-farming population, are employed in the gig economy, which is thought to contribute 1.25% of the country’s GDP [8]. But the gig economy has drawbacks as well. Gig employees have more leeway in the sense of work and hours, more autonomy in conducting assignments, more gig options, and the potential for higher earnings via commissions. However, they also face difficulties, such as increased stress levels and less financial stability due to an inability to deduct work-related costs from their income. While the advantages to enterprises of the gig economy are clear—lower operational expenses (pay per job), rapid scalability, and access to a wide pool of flexible workers—the gig economy also has drawbacks, including a less dependable workforce and stringent controls on contract status, among others [9]. The demand and supply figures may cause these benefits and drawbacks to shift for companies and employees. To augment their income, many people in India, where the unemployment rate is fairly high, use ridesharing services, like Uber and Ola. Even though there are more employees, businesses are paying them less, even though they are making them work longer hours every day. Workers’ health might suffer from the added stress, and accidents can happen more often because employees are fidgety. In addition, as the laborers are not considered employees, the firms involved cannot be held responsible for any damages that may occur [1]. However, this situation only applies to low-skilled jobs, which abound, despite market constraints. Professional jobs and contracts, on the other hand, are in great demand, but in short supply, so freelancers in these fields may expect high wages. Therefore, employers have more power to make choices in low-skilled jobs, whereas gig workers have more power in professional services. The usage of the gig economy in the healthcare field is one intriguing subject for further research. For instance, Saudi Arabia’s healthcare system is quickly changing as a result of the Vision 2030. However, the country’s healthcare industry depends largely on foreign labor. The majority of healthcare workers in Saudi Arabia are nurses, and they are essential to managing the daily operations of several hospitals there, i.e., they are the backbone of healthcare. The 2018 ratio of 5.5 nurses per 1000 persons in Saudi Arabia is above average when compared with other MENA countries [10,11,12]. However, there has been an increase in the use of foreign nurses in recent years. Only 42.9% of Saudi Arabia’s 196,701 registered nurses were Saudi residents in 2020; the remaining 57.1% were foreigners. The rising prevalence of chronic illnesses in Saudi Arabia means a greater need for nurses in the coming years. There has been no noticeable shift in the nursing workforce in the twenty years since the government issued a royal order to support a Saudization strategy for the nursing profession in 1992 [13,14]. Several factors have been identified as barriers to Saudi nationals entering the nursing workforce. The nursing field lacks gender segregation [15,16], female nurses caring for male patients sometimes have no family support [17,18], many nurses are underpaid and put in long hours [15,18], and there are few prospects for growth [10,19]. Therefore, the urgent demand for contract employees and freelancers is significant in Saudi Arabia. Recent difficulties with obtaining work permits for expatriates have been tied mostly to the rise of contract workers. In addition, the elimination of the Kafala sponsorship system (in which an employee’s visa status is dependent on their employer) grants foreign nationals more adaptability in terms of where they work and when they enter and leave the country [20]. eHealth services, both those provided remotely and those delivered in hospitals, have been the primary focus of research into the potential for the gig economy’s integration into the Saudi Arabian healthcare system as part of digitalization [21,22,23,24]. The gig economy’s usefulness and advantages have been examined from a number of angles, and the findings are encouraging and persuasive. Sustainable healthcare development, motivation for the Saudization program, economic growth through job creation [21], improved access via remote services, efficient delivery of healthcare services [22,23,24], better doctor–patient relationships [25,26], etc., are just some of the many areas where the gig economy has proven its worth. The implementation of the gig economy and the variables influencing alignment with the GEF were studied in [27]. Using public data on drivers’ outside alternatives and private data from a large ride-hailing platform on individual drivers’ particular travels, the research cited in [28] examined gig workers’ dynamic multihoming decisions. They used a combination of simulation and machine learning to assess the hidden costs that shape drivers’ foresightful choices and actions. Their research showed that employees are short-sighted and prefer stable compensation over sporadic bonuses. Using this information, they conducted counterfactual assessments to look at how different pay and incentive structures might affect the platform’s workforce and productivity. They realized that constant compensation is more effective than variable pay per job at keeping multihoming employees on the platform. Incentives for continuous employment or the postponement of resignations are two more ways by which platforms might regulate their access to workers. Their study analyzed the effects of New York City’s 2018 Driver Income Rules on multihoming behavior, providing important insights for gig economy platforms and regulators in the formulation of appropriate incentive schemes and laws. The purpose of this study [29] was to provide a comprehensive analysis of the factors that play a role in freelancers’ financial decisions. This was a quantitative study based on responses from 410 participants. The participants in the research were members of the gig economy in the Indonesian city of Bandung. Investment expertise, risk aversion, profitability, and fiscal literacy were all studied for their impact on stock picking. The findings shed light on many critical factors that influence investment decisions in the gig economy. Study [30] used gravitational models to examine how geographical, institutional, and cultural distance influenced over 30,000 platform hires across 26 European nations. Gig workers are still preferentially employed from economically similar regions; therefore, the online gig economy is not limitless, despite the fact that these platforms are utilized to off-shore labor from high- to low-wage nations. Having a similar language also makes it easier to hire people from other nations. However, employment practices vary little from country to country, which suggests that online platforms develop their own institutional framework, independent of the official and informal structures in place. We draw the conclusion that the online gig economy is not an unlimited nor a frictionless labor market, suggesting that claims that it would make employment options more accessible to everybody are overstated. In order to identify the factors that prevent gig workers from joining gig platforms, the researchers in [31] used an interpretative structural model (ISM). Along with MICMAC analysis, it confirmed the connection and elucidated the causes of the correlation. The design process included input from gig workers and specialists affiliated with food delivery gig platforms, like Zomato and Swiggy. The purpose of the study [32] was to examine the effects of gender, age, and level of education on freelancers’ subjective well-being. An online questionnaire was used to assess 471 independent contractors’ feelings of satisfaction with their lives in the “gig” economy in Serbia, Bosnia and Herzegovina, Macedonia, and Montenegro. Two distinct clusters were found using factor analysis as the principal statistical technique—the effects of freelancing on one’s life and health outside of work, and whether or not one’s financial and professional goals are met. Overall job satisfaction did not differ significantly by gender. However, older freelancers who had been in the workforce longer reported higher levels of satisfaction with regard to the realization of economic and professional goals. A critical gap that has been identified in previous studies is the lack of use of artificial neural networks in the analysis and predictions of target parameters in the field of the gig economy.

Considering the recent developments in the field of artificial neural networks and the many capabilities of this powerful tool, in this research, we decided to use artificial neural networks to predict the alignment of employees with the gig economy framework. Excellent data were gathered and shared by them. Using the results of previous studies [27], this investigation aimed to develop a neural network that could be used to make predictions about respondents’ compatibility with the GEF by analyzing their responses to demographic and socioeconomic surveys. We will begin with a thorough presentation of the data-gathering procedures and questionnaire questions. The information is fed into a multilayer perceptron (MLP) neural network in the next section. The last two parts report the findings of the neural network and the overall conclusions. The general trend of the current article can be seen in Figure 1.

2. Data Acquisition

In [27], a cross-sectional survey was used to gather information on the nursing industry’s use of the GEF. Since the data from [27] were used in this investigation, we shall briefly discuss the aforementioned surveys.

The survey asked questions specific to the nursing profession in Saudi Arabia and the GEF, with attention being paid to other relevant topics. There are two sections in the questionnaire. In the first section, there are six questions on the demographics of the study’s subjects. The 15 primary survey questions are discussed in the second section. The first inquiry pertains to the difficulties encountered by Saudi nationals in entering the nursing profession, as discussed in Ref. [10]. Responses to Questions 2–9 are gathered to represent the features of the GEF and their relevance to nurse management, as outlined in Ref. [33]. The disadvantages of the GEF for nurse management are discussed in questions 10 and 11, and the framework’s influence on the motivation of Saudi nationals is evaluated. The viability of incorporating the GEF into the Saudization and Vision 2030 programs is the topic of questions 12 and 13, while the requirement for legislative restrictions for applying the GEF to the nursing profession is the focus of question 14. Questions 1–14 were answered using a five-point Likert scale, with 1 being strong disagreement, 2 being disagreement, 3 being neutral, 4 being agreement, and 5 representing strong agreement. The 15th question asks about the life, personal, professional, and financial impacts of gig economy participation for nurses. Participants can score each item in question 15 a rating on a five-point Likert scale, with one, two, three, four, and five representing “not at all”, “slightly”, “moderately”, “and “highly” advantageous, respectively, all of which reflect positive aspects of the gig economy. The pilot research with 35 nurses helped to assess the questionnaire’s validity. Cronbach’s alpha was attained in the preliminary investigation. Good internal dependability and consistency were achieved (alpha > 0.70 across the board) [34]. The survey’s intended audience was Saudi Arabian hospital nurses; thus, the link was shared with human resources departments at 102 facilities across the country. This included 82 government hospitals and 20 private hospitals. Cochran’s method was used to determine a sample size based on the total number of registered nurses in Saudi Arabia (196,701) [10], and the results indicated a sample size of 383. There were a total of 406 replies. In all, 379 answers were used for analysis after missing data were removed [35]. SPSS version 20.0 was used to perform statistical analyses, such as t-tests, mean comparisons, and standard deviation calculations. In order to prevent the findings from being skewed, missing data were omitted. Because the survey contained questions with a variety of answer choices, such as rating scales for the suitability of characteristics of the gig economy to healthcare and multiple-choice options for choosing challenges, the data analysis is shown in a variety of percentages to indicate the challenges, as well as in means and standard deviations to illustrate the practicality and effect of the gig economy on medical services. The nursing field was found to be the best fit for gig economy traits, like job flexibility (mean = 4.3), work freedom (mean = 4.2), and duty variety (mean = 4.1). The gig economy has different effects on daily life, business development, and personal growth in different ways, and these disparities were statistically significant (p < 0.05). Participants who were women thought that working in the gig economy was better for their personal and professional growth than participants who were men. Compared with older adults (mean = 3.7, SD = 1.97), people under 40 who were younger (mean = 4.1, SD = 1.24) had a large impact [27]. Before taking the survey, each participant checked a box to show that they understood the goal and rules of the study. During the whole study, the subjects’ privacy will be kept safe. The Saudi Arabian Ministry of Health’s ethics group gave the go-ahead for the study. The study revealed several challenges faced by the nursing profession in Saudi Arabia. These included gender-mixed workplaces, cultural and socioeconomic obstacles for Saudi nurses, lower salaries, unhappiness among family members when their loved ones care for patients of the opposite gender, a heavy workload, and an increased reliance on expatriates.

The elements of the GEF are well-suited to overcoming these obstacles. The participants selected ‘flexibility of work’ as useful in integrating the gig economy into the nursing profession. This allows nurses to pick and choose which hospitals they want to work at. Both Saudi female nurses looking for employment in local hospitals and expat nurses looking to work at hospitals of their interest might profit from this. Freedom of assignment, for example, lets nurses choose shifts working with patients of the same or opposite gender. This allows nurses to lessen the burden of family involvement in patient care. The nurse’s professional and personal lives may both benefit from contract employment because of the flexibility it provides. In addition, nurses may choose hours and facilities that fit best for them via gig platforms, increasing their own responsibility for their job and the care they provide. In addition, nurses may find it simpler to take on contract-based duties that can be performed remotely using these platforms. This includes tasks like medication reminders, monitoring patient progress, and keeping nursing records.

Previous research [21,22,23,24] has examined the feasibility of gig labor in healthcare settings generally, but not in relation to the nursing profession. Negatives associated with the GEF, such as a lack of benefits and unpredictable compensation, were shown to have little effect on nurses’ enthusiasm for the profession. Increases in the number of Saudi nursing employees, especially among women, and efficient management of nursing activities through gig platforms are two examples of the benefits that can help achieve the goals of Saudization and Vision 2030 in the realm of medical care. Unfortunately, while the gig Economy is still in its infancy, no comprehensive laws or regulations have yet been drafted, heightening the danger faced by both gig workers and the firms that hire them. As a result, the majority of respondents felt that the nursing profession may benefit from the creation and implementation of legislative rules and policies to facilitate the use of the GEF by healthcare providers and workers, like nurses. A future study may also address the dearth of literature on the topic of planning and implementing governmental rules for healthcare gig economy adoption [27]. For the answers to the surveys, which are qualitative, to be used in the creation of an artificial neural network, they must first be turned into quantitative data.

Table A1. The questions asked of the participants and their numerical equivalents.

Questions	Numerical Equivalents
Gender:	Male = 1 Female = 2
Age:	18–20 = 1 21–30 = 2 31–40 = 3 41–50 = 4 51–60 = 5 >60 = 6
Education:	High school = 1, Diploma = 2 Bachelor’s degree = 3 Master’s degree = 4 Ph.D. = 5 Others = 6
Role:	Staff nurse = 1 Nurse manager = 2
Experience:	<3 years = 1 3–6 years = 2 7–9 years = 3 10–12 years = 4 >12 years = 5
Nationality:	Saudi = 1 Non-Saudi = 2
Please mark off the items below that have an impact on Saudi Arabia’s nursing industry.	➢ Greater dependency on expatriates = 1 ➢ Gender-mixed working environment = 2 ➢ Poor patient communication because foreign nurses do not understand Saudi culture =3 ➢ Retaining talented nurses = 4 ➢ Costs rising as a result of reliance on foreign workers = 5 ➢ Families do not like it when their children or spouses look after patients of the opposite gender at hospitals = 6 ➢ High workload = 7 ➢ Negative public perception of nursing in Saudi Arabia = 8 ➢ Nursing professionals’ heavy schedule prevents them from spending time with their families = 9 ➢ Saudi nurses are subject to sociocultural limitations = 10 ➢ Convenient work hours include weekends, night shifts, and public holidays = 11 ➢ Lack of professional growth = 12 ➢ Lower wages = 13 ➢ Others = 14

in Appendix A displays the inquiries posed to the participants and their corresponding numerical representations.

MLP Neural Network

Neurons serve as the fundamental structural units of the brain. The neural structure comprises numerous neural units, which communicate with each other via dendritic extensions. Upon the completion of intracellular processes, the axon transmits information to the external environment. In the past, these processes have been studied as part of physiology and biochemistry. However, scientists’ use of mathematical modeling has changed the field in a major way. Numerous studies attest to the growing interest of academics in the use of sophisticated mathematical methods and artificial neural networks in a wide range of scientific disciplines [36,37,38]. The multilayer perceptron (MLP) neural network is often used in modeling. MATLAB version 2018b was used to create the MLP neural networks and extract the aforementioned properties. In this research, neural networks were not generated using pre-made toolboxes, but rather were painstakingly hand-coded, from training to validation to testing. This MATLAB package may include a number of toolboxes for building neural networks. It is important to note that the neural network was trained using the “feedforwardnet” function. The input layer, hidden layer, and output layer are the three main parts of the network’s structure. Neural networks may make use of several hidden layers. Within the concealed layers, activation functions are implemented to carry out the necessary mathematical processes. The qualities and level of non-linear properties presented by the available data will determine the number of layers, the number of neurons in the hidden layer, and the selection of the activation function. Neuronal output may be represented as [39,40]:

$${\mathrm{n}}_{\mathrm{l}}={\sum }_{\mathrm{i}=1}^{\mathrm{u}}{\mathrm{x}}_{\mathrm{i}}{\mathrm{w}}_{\mathrm{ij}}+\mathrm{b} \mathrm{j}=1,2,\cdots ,\mathrm{m}$$

(1)

$${\mathrm{u}}_{\mathrm{j}}=\mathrm{f}\left({\sum }_{\mathrm{i}=1}^{\mathrm{u}}{\mathrm{x}}_{\mathrm{i}}{\mathrm{w}}_{\mathrm{ij}}+\mathrm{b}\right) \mathrm{j}=1,2,\cdots ,\mathrm{m}$$

(2)

$$\mathrm{output}={\sum }_{\mathrm{n}=1}^{\mathrm{j}}\left({\mathrm{u}}_{\mathrm{n}}{\mathrm{w}}_{\mathrm{n}}\right)+\mathrm{b}$$

(3)

where x is shorthand for the set of inputs. In this context, the parameters of interest are the weighting factor W, the bias term b, and the activation function f. The “i” values represent the input, while the “j” values represent the number of hidden layer neurons currently being used. In the design of the neural network in this research, the number of hidden layers and the number of neurons in the hidden layer in a learning and evaluation process varied from 1 hidden layer to 5 hidden layers and the number of neurons in each layer varied from 5 neurons to 40 neurons. After designing different neural networks, it was found that a network with 3 hidden layers and 30, 28, and 15 hidden-layer neurons, respectively, had the best performance in predicting the target outputs. To prevent overfitting and underfitting, the data were separated into three distinct sets: training, validation, and testing. A proportion of 70% of the data was allocated to training, 15% to validation, and 15% to the final test. It should be noted that this division was completely random. The observation and fitting procedures of the neural network rely heavily on the training dataset. “Validation data” usually refers to a part of the dataset that is used to measure how well the training process is working. During training, these data are used to judge how reliable the networks are. Validating the performance of the trained neural network requires the use of test data. A neural network has to be proficient with all three kinds of inputs for it to succeed in the real world.

3. Result and Discussion

In this research, an MLP neural network was used to determine the relationship between demographic parameters and survey responses by aligning with the GEF. The neural network utilized 20 inputs, encompassing demographic parameters, such as gender, age, education, role, work experience, and nationality, in addition to responses to 14 survey questions. The network’s output pertained to the inquiry regarding the influence of the fig economy on four domains of individual life, namely personal life, personal development, professional development, and financial management. All these data were extracted from research [27]. A total of 379 samples were available for this research, of which 267 were used for training the neural network, 57 for validation, and 57 for the final test. The available data were qualitative data that were converted into quantitative data to be used as the inputs and outputs of the neural network. The output data include five different classes. Classes 1 to 5 refer to “not at all advantageous” to “highly advantageous”, respectively. The present investigation involved the implementation and examination of various neural networks, ranging from one to five hidden layers, and varying in the number of neurons within the hidden layer. Figure 2 depicts the configuration of the aforementioned network.

The neural network that was developed demonstrated a capacity to accurately classify outputs ranging from 1 to 4, with a notable degree of precision, specifically 96.5%, 96.5%, 99.2%, and 99.2%, respectively. The performance of this network for each output is shown in Figure 3, Figure 4, Figure 5 and Figure 6 using the confusion matrix for three categories of training, validation, and testing data. In this matrix, the number and percentage of correct and incorrect answers of the neural network can be seen separately for each class. For a better understanding of this figure, for example, in Figure 3c, all the data are correctly classified by the neural network, except for one datum that was related to the data of class 3, which was wrongly classified into class 5. This study demonstrated that a neural network may be used to accurately forecast how well individuals would fit with the aims of the gig economy. The designed neural network can be used in Saudi Arabian hospitals to predict the alignment of their nurses with the gig economy framework. It can also be used in nurse recruitment systems to assess individuals to determine whether they align with this framework or not. The methodology presented in the current research can be used in many companies to improve efficiency and also to examine the impact of the GEF. Also, by examining different people in various companies and using different neural networks, it is possible to predict and examine the alignment of company employees with the GEF. Researchers in this field can investigate different neural networks and different methods of feature selection in future research so that they can achieve higher accuracy and reduce the number of calculations by selecting effective features. One of the limitations of the current research is the study of the population limited to Saudi Arabian hospitals. For the development of the current research, different statistical communities can be investigated and, by using artificial neural networks, the relationship between different parameters and the compatibility with the gig economy framework can be found.

4. Conclusions

One of the cornerstones of any healthcare system is its nursing workforce, and this research assessed the applicability of the GEF to the nursing workforce in Saudi Arabia. The flexibility and adaptability afforded by the gig economy may help Saudi Arabia’s nursing profession overcome some of the most intractable problems impeding the country’s progress toward its Saudization and Vision 2030 healthcare goals. In this research, an artificial neural network was used to find the relationship between demographic characteristics and some questionnaires by aligning with the GEF. The designed network had 20 inputs and 4 outputs. The outputs of this network included the impact of the gig economy on life, professional development, personal development, and financial management, which the participants in the survey answered on five levels, from not at all advantageous to highly advantageous. The network outputs were able to predict the alignment with the GEF in the four mentioned aspects with accuracies of 96.5%, 99.2%, 96.5%, and 99.2%, respectively. The neural network presented in this research can be used in hospitals to examine nurses in terms of alignment with GEF. Also, different recruiting companies can use the neural network provided in this research to evaluate the compatibility of candidates with the gig economy. The development of the current research with different statistical communities can help to provide a reliable and comprehensive method to predict alignment with the gig economy.